refactor(workflows): complete skills→workflows rename (internal + wire/FE) [#3324 follow-up] (#3412)

Co-authored-by: Claude <noreply@anthropic.com>

app/src/components/intelligence/WorkflowsTab.tsx

@@ -5,7 +5,7 @@
  * The Intelligence page's "Workflows" tab — the single home for installed
  * workflows (the unified primitive: a goal + the procedure to reach it,
  * authored as SKILL.md bundles and served by the `workflows_*` JSON-RPC via
- * `skillsApi`).
+ * `workflowsApi`).
  *
  * Owns the full workflow surface that used to live on the Connections page:
  *   - lists discovered workflows as cards,
@@ -21,7 +21,7 @@ import { useCallback, useEffect, useState } from 'react';
 import { useNavigate } from 'react-router-dom';
 
 import { useT } from '../../lib/i18n/I18nContext';
-import { skillsApi, type SkillSummary } from '../../services/api/skillsApi';
+import { workflowsApi, type WorkflowSummary } from '../../services/api/workflowsApi';
 import type { ToastNotification } from '../../types/intelligence';
 import CreateSkillModal from '../skills/CreateSkillModal';
 import UnifiedSkillCard from '../skills/SkillCard';
@@ -34,10 +34,10 @@ const log = debug('intelligence:workflows');
 export default function WorkflowsTab() {
   const { t } = useT();
   const navigate = useNavigate();
-  const [workflows, setWorkflows] = useState<SkillSummary[]>([]);
+  const [workflows, setWorkflows] = useState<WorkflowSummary[]>([]);
   const [loading, setLoading] = useState(true);
   const [createModalOpen, setCreateModalOpen] = useState(false);
-  const [uninstallCandidate, setUninstallCandidate] = useState<SkillSummary | null>(null);
+  const [uninstallCandidate, setUninstallCandidate] = useState<WorkflowSummary | null>(null);
   const [loadError, setLoadError] = useState<string | null>(null);
   const [toasts, setToasts] = useState<ToastNotification[]>([]);
 
@@ -49,9 +49,9 @@ export default function WorkflowsTab() {
     setToasts(prev => prev.filter(toast => toast.id !== id));
   }, []);
 
-  const refresh = useCallback(async (): Promise<SkillSummary[]> => {
+  const refresh = useCallback(async (): Promise<WorkflowSummary[]> => {
     try {
-      const list = await skillsApi.listSkills();
+      const list = await workflowsApi.listWorkflows();
       log('listWorkflows ok count=%d', list.length);
       setLoadError(null);
       setWorkflows(list);
@@ -100,7 +100,7 @@ export default function WorkflowsTab() {
         </div>
       </div>
 
-      {/* Load error — shown instead of the empty state when listSkills fails,
+      {/* Load error — shown instead of the empty state when listWorkflows fails,
           so an outage doesn't read as "you have no workflows". */}
       {loadError && !loading ? (
         <div

app/src/components/intelligence/__tests__/WorkflowsTab.test.tsx

@@ -1,7 +1,7 @@
 import { fireEvent, render, screen, waitFor } from '@testing-library/react';
 import { beforeEach, describe, expect, it, vi } from 'vitest';
 
-import type { SkillSummary } from '../../../services/api/skillsApi';
+import type { WorkflowSummary } from '../../../services/api/workflowsApi';
 import WorkflowsTab from '../WorkflowsTab';
 
 vi.mock('../../../lib/i18n/I18nContext', () => ({ useT: () => ({ t: (k: string) => k }) }));
@@ -18,7 +18,7 @@ vi.mock('../../skills/CreateSkillModal', () => ({
     onCreated,
     onClose,
   }: {
-    onCreated: (wf: SkillSummary) => void;
+    onCreated: (wf: WorkflowSummary) => void;
     onClose: () => void;
   }) => (
     <div data-testid="create-modal-stub">
@@ -62,7 +62,7 @@ vi.mock('../../skills/UninstallSkillConfirmDialog', () => ({
   ),
 }));
 
-const seeded = (overrides: Partial<SkillSummary>): SkillSummary => ({
+const seeded = (overrides: Partial<WorkflowSummary>): WorkflowSummary => ({
   id: 'wf-1',
   name: 'WF 1',
   description: 'A workflow.',
@@ -82,22 +82,25 @@ const seeded = (overrides: Partial<SkillSummary>): SkillSummary => ({
   ...overrides,
 });
 
-const listSkills = vi.fn();
-vi.mock('../../../services/api/skillsApi', async () => {
-  const actual = await vi.importActual<typeof import('../../../services/api/skillsApi')>(
-    '../../../services/api/skillsApi'
+const listWorkflows = vi.fn();
+vi.mock('../../../services/api/workflowsApi', async () => {
+  const actual = await vi.importActual<typeof import('../../../services/api/workflowsApi')>(
+    '../../../services/api/workflowsApi'
   );
-  return { ...actual, skillsApi: { ...actual.skillsApi, listSkills: () => listSkills() } };
+  return {
+    ...actual,
+    workflowsApi: { ...actual.workflowsApi, listWorkflows: () => listWorkflows() },
+  };
 });
 
 describe('WorkflowsTab', () => {
   beforeEach(() => {
     navigate.mockReset();
-    listSkills.mockReset();
+    listWorkflows.mockReset();
   });
 
-  it('lists workflows from skillsApi with the create entry point', async () => {
-    listSkills.mockResolvedValue([
+  it('lists workflows from workflowsApi with the create entry point', async () => {
+    listWorkflows.mockResolvedValue([
       seeded({ id: 'user-wf', name: 'User WF', scope: 'user' }),
       seeded({ id: 'project-wf', name: 'Project WF', scope: 'project' }),
     ]);
@@ -115,7 +118,7 @@ describe('WorkflowsTab', () => {
   });
 
   it('navigates to the locked runner page when a card is opened', async () => {
-    listSkills.mockResolvedValue([seeded({ id: 'user-wf', name: 'User WF', scope: 'user' })]);
+    listWorkflows.mockResolvedValue([seeded({ id: 'user-wf', name: 'User WF', scope: 'user' })]);
     render(<WorkflowsTab />);
     await waitFor(() => expect(screen.getByTestId('workflow-open-user-wf')).toBeInTheDocument());
 
@@ -124,7 +127,7 @@ describe('WorkflowsTab', () => {
   });
 
   it('opens create and lands on the new workflow runner after create', async () => {
-    listSkills.mockResolvedValue([]);
+    listWorkflows.mockResolvedValue([]);
     render(<WorkflowsTab />);
     await waitFor(() => expect(screen.getByTestId('workflows-create-btn')).toBeInTheDocument());
 
@@ -133,8 +136,8 @@ describe('WorkflowsTab', () => {
     expect(navigate).toHaveBeenCalledWith('/workflows/run?workflow=new-wf&lock=1');
   });
 
-  it('shows an error panel with retry when listSkills fails (not the empty state)', async () => {
-    listSkills.mockRejectedValueOnce(new Error('backend down'));
+  it('shows an error panel with retry when listWorkflows fails (not the empty state)', async () => {
+    listWorkflows.mockRejectedValueOnce(new Error('backend down'));
     render(<WorkflowsTab />);
 
     await waitFor(() => expect(screen.getByTestId('workflows-load-error')).toBeInTheDocument());
@@ -142,13 +145,15 @@ describe('WorkflowsTab', () => {
     expect(screen.queryByText('workflows.empty.title')).not.toBeInTheDocument();
 
     // Retry re-fetches and renders the list.
-    listSkills.mockResolvedValueOnce([seeded({ id: 'user-wf', name: 'User WF', scope: 'user' })]);
+    listWorkflows.mockResolvedValueOnce([
+      seeded({ id: 'user-wf', name: 'User WF', scope: 'user' }),
+    ]);
     fireEvent.click(screen.getByText('common.retry'));
     await waitFor(() => expect(screen.getByText('User WF')).toBeInTheDocument());
   });
 
   it('removes the workflow from the list after uninstall, keyed by id', async () => {
-    listSkills.mockResolvedValue([seeded({ id: 'user-wf', name: 'User WF', scope: 'user' })]);
+    listWorkflows.mockResolvedValue([seeded({ id: 'user-wf', name: 'User WF', scope: 'user' })]);
     render(<WorkflowsTab />);
     await waitFor(() => expect(screen.getByTestId('workflow-card-user-wf')).toBeInTheDocument());
 
@@ -162,7 +167,7 @@ describe('WorkflowsTab', () => {
   });
 
   it('renders the empty state when no workflows are installed', async () => {
-    listSkills.mockResolvedValue([]);
+    listWorkflows.mockResolvedValue([]);
     render(<WorkflowsTab />);
     await waitFor(() => expect(screen.getByText('workflows.empty.title')).toBeInTheDocument());
     expect(screen.queryByTestId('workflows-list')).not.toBeInTheDocument();

app/src/components/skills/CreateSkillModal.tsx

@@ -20,7 +20,7 @@ import { useCallback, useEffect, useRef, useState } from 'react';
 import { createPortal } from 'react-dom';
 
 import { useT } from '../../lib/i18n/I18nContext';
-import { type SkillSummary } from '../../services/api/skillsApi';
+import { type WorkflowSummary } from '../../services/api/workflowsApi';
 import CreateWorkflowForm from './CreateWorkflowForm';
 
 const log = debug('skills:create-modal');
@@ -29,9 +29,9 @@ const CREATE_FORM_ID = 'create-skill-modal-form';
 
 interface Props {
   onClose: () => void;
-  onCreated: (skill: SkillSummary) => void;
+  onCreated: (skill: WorkflowSummary) => void;
   /** When set, the modal edits this workflow instead of creating a new one. */
-  editing?: SkillSummary;
+  editing?: WorkflowSummary;
 }
 
 export default function CreateSkillModal({ onClose, onCreated, editing }: Props) {

app/src/components/skills/CreateWorkflowForm.tsx

@@ -9,7 +9,7 @@
  *   - All form fields (name, description, scope, license, author,
  *     tags, allowed-tools).
  *   - Slug preview + validation (name and description required).
- *   - Submit handler that calls `skillsApi.createSkill` and surfaces
+ *   - Submit handler that calls `workflowsApi.createWorkflow` and surfaces
  *     the result via `onCreated(skill)` / error string via inline
  *     `<div role="alert">`.
  *
@@ -41,12 +41,12 @@ import {
 
 import { useT } from '../../lib/i18n/I18nContext';
 import {
-  type CreateSkillInput,
-  type CreateSkillInputDef,
-  type SkillScope,
-  type SkillSummary,
-  skillsApi,
-} from '../../services/api/skillsApi';
+  type CreateWorkflowInput,
+  type CreateWorkflowInputDef,
+  type WorkflowScope,
+  type WorkflowSummary,
+  workflowsApi,
+} from '../../services/api/workflowsApi';
 
 /** Mirrors `SkillCreateInputDef` shape used as wire payload, with one
  *  extra `localId` for stable React keys across re-renders (the wire
@@ -80,7 +80,7 @@ const log = debug('skills:create-form');
 export interface CreateSkillFormHandle {
   /** True iff name+description are present and no submit is in flight. */
   isValid: () => boolean;
-  /** True while skillsApi.createSkill is in flight. */
+  /** True while workflowsApi.createWorkflow is in flight. */
   isSubmitting: () => boolean;
   /** Imperatively trigger submit. Resolves once the round-trip finishes. */
   submit: () => Promise<void>;
@@ -94,7 +94,7 @@ export interface CreateSkillFormProps {
    */
   formId: string;
   /** Called with the freshly-created skill on success. */
-  onCreated: (skill: SkillSummary) => void;
+  onCreated: (skill: WorkflowSummary) => void;
   /**
    * Called whenever validity / submission state changes so the
    * wrapper can sync its submit button's disabled state without
@@ -110,7 +110,7 @@ export interface CreateSkillFormProps {
    * allowed-tools (not exposed as editable fields) are carried through so
    * they're preserved on save.
    */
-  editing?: SkillSummary;
+  editing?: WorkflowSummary;
 }
 
 /**
@@ -156,7 +156,7 @@ const CreateWorkflowForm = forwardRef<CreateSkillFormHandle, CreateSkillFormProp
     // dashboard-created skills. Project-scoped skills are still
     // creatable by editing the workspace skill files directly. The
     // backend payload still requires `scope` so we hold it as a const.
-    const scope: SkillScope = 'user';
+    const scope: WorkflowScope = 'user';
     const [submitting, setSubmitting] = useState(false);
     const [error, setError] = useState<string | null>(null);
     const [inputs, setInputs] = useState<InputRow[]>([]);
@@ -217,7 +217,7 @@ const CreateWorkflowForm = forwardRef<CreateSkillFormHandle, CreateSkillFormProp
       let cancelled = false;
       void (async () => {
         try {
-          const desc = await skillsApi.describeSkill(editing.id);
+          const desc = await workflowsApi.describeWorkflow(editing.id);
           if (cancelled) return;
           if (desc.when_to_use) setWhenToUse(desc.when_to_use);
           setInputs(
@@ -243,7 +243,7 @@ const CreateWorkflowForm = forwardRef<CreateSkillFormHandle, CreateSkillFormProp
 
     const submit = useCallback(async () => {
       if (!formValid) return;
-      const payload: CreateSkillInput = {
+      const payload: CreateWorkflowInput = {
         name: name.trim(),
         description: description.trim(),
         scope,
@@ -264,8 +264,8 @@ const CreateWorkflowForm = forwardRef<CreateSkillFormHandle, CreateSkillFormProp
         }
       }
       if (inputs.length > 0) {
-        payload.inputs = inputs.map<CreateSkillInputDef>((r) => {
-          const def: CreateSkillInputDef = {
+        payload.inputs = inputs.map<CreateWorkflowInputDef>((r) => {
+          const def: CreateWorkflowInputDef = {
             name: r.name.trim(),
             required: r.required,
           };
@@ -282,8 +282,8 @@ const CreateWorkflowForm = forwardRef<CreateSkillFormHandle, CreateSkillFormProp
       setError(null);
       try {
         const saved = editing
-          ? await skillsApi.updateSkill(payload)
-          : await skillsApi.createSkill(payload);
+          ? await workflowsApi.updateWorkflow(payload)
+          : await workflowsApi.createWorkflow(payload);
         log('submit-ok id=%s edit=%s', saved.id, isEdit);
         onCreated(saved);
       } catch (err) {

app/src/components/skills/InstallSkillDialog.tsx

@@ -16,7 +16,7 @@
  *     optional timeout in seconds.
  *   - While the RPC is in flight we show a "Fetching…" indicator and
  *     disable close / backdrop-dismiss so the caller sees the outcome.
- *   - On success we surface the list of `newSkills` (ids that appeared
+ *   - On success we surface the list of `newWorkflows` (ids that appeared
  *     post-install) plus captured fetch log / parse-warning panes, then
  *     hand the result back to the caller via `onInstalled` so the
  *     parent can refetch the list and auto-select the row.
@@ -34,10 +34,10 @@ import debug from 'debug';
 
 import { useT } from '../../lib/i18n/I18nContext';
 import {
-  skillsApi,
-  type InstallSkillFromUrlResult,
-  type SkillSummary,
-} from '../../services/api/skillsApi';
+  workflowsApi,
+  type InstallWorkflowFromUrlResult,
+  type WorkflowSummary,
+} from '../../services/api/workflowsApi';
 import { trackEvent } from '../../services/analytics';
 
 const log = debug('skills:install-dialog');
@@ -47,17 +47,17 @@ interface Props {
   /**
    * Fires when the backend reports the install succeeded. The parent is
    * responsible for refetching the skills list (the RPC already returns
-   * the freshly-added ids, but the caller may want full `SkillSummary`
-   * rows). `newSkills` lists ids that appeared post-install.
+   * the freshly-added ids, but the caller may want full `WorkflowSummary`
+   * rows). `newWorkflows` lists ids that appeared post-install.
    */
-  onInstalled: (result: InstallSkillFromUrlResult) => void;
+  onInstalled: (result: InstallWorkflowFromUrlResult) => void;
   /**
    * Optional: used only for symmetry with `CreateSkillModal`. When
    * supplied and the caller wants to auto-open the detail drawer for a
-   * specific skill, they can resolve the full `SkillSummary` and call
+   * specific skill, they can resolve the full `WorkflowSummary` and call
    * this directly. Not invoked by the dialog itself.
    */
-  onSelectSkill?: (skill: SkillSummary) => void;
+  onSelectSkill?: (skill: WorkflowSummary) => void;
 }
 
 /**
@@ -151,7 +151,7 @@ export default function InstallSkillDialog({ onClose, onInstalled }: Props) {
   const [timeoutSecs, setTimeoutSecs] = useState<string>('');
   const [submitting, setSubmitting] = useState(false);
   const [error, setError] = useState<string | null>(null);
-  const [result, setResult] = useState<InstallSkillFromUrlResult | null>(null);
+  const [result, setResult] = useState<InstallWorkflowFromUrlResult | null>(null);
 
   const firstFieldRef = useRef<HTMLInputElement | null>(null);
   const previousFocusRef = useRef<HTMLElement | null>(null);
@@ -201,14 +201,14 @@ export default function InstallSkillDialog({ onClose, onInstalled }: Props) {
       setSubmitting(true);
       setError(null);
       try {
-        const installed = await skillsApi.installSkillFromUrl(payload);
+        const installed = await workflowsApi.installWorkflowFromUrl(payload);
         log(
           'submit-ok new=%d stdout=%d stderr=%d',
-          installed.newSkills.length,
+          installed.newWorkflows.length,
           installed.stdout.length,
           installed.stderr.length
         );
-        for (const skillId of installed.newSkills) {
+        for (const skillId of installed.newWorkflows) {
           trackEvent('skill_install', { skill_id: skillId });
         }
         setResult(installed);
@@ -383,16 +383,16 @@ export default function InstallSkillDialog({ onClose, onInstalled }: Props) {
                 <div>
                   <p className="font-semibold">{t('skills.install.installComplete')}</p>
                   <p className="mt-1">
-                    {result.newSkills.length > 0
+                    {result.newWorkflows.length > 0
                       ? t('skills.install.successDiscovered').replace(
                           '{count}',
-                          String(result.newSkills.length)
+                          String(result.newWorkflows.length)
                         )
                       : t('skills.install.successNoNewIds')}
                   </p>
-                  {result.newSkills.length > 0 ? (
+                  {result.newWorkflows.length > 0 ? (
                     <ul className="mt-1 list-disc pl-5 font-mono">
-                      {result.newSkills.map(id => (
+                      {result.newWorkflows.map(id => (
                         <li key={id}>{id}</li>
                       ))}
                     </ul>

app/src/components/skills/SkillsExplorerTab.tsx

@@ -7,10 +7,10 @@ import {
   type CatalogEntry,
 } from '../../services/api/skillRegistryApi';
 import {
-  skillsApi,
-  type InstallSkillFromUrlResult,
-  type SkillSummary,
-} from '../../services/api/skillsApi';
+  workflowsApi,
+  type InstallWorkflowFromUrlResult,
+  type WorkflowSummary,
+} from '../../services/api/workflowsApi';
 import EmptyStateCard from '../EmptyStateCard';
 import InstallSkillDialog from './InstallSkillDialog';
 import UninstallSkillConfirmDialog from './UninstallSkillConfirmDialog';
@@ -67,7 +67,7 @@ function SourceBadge({ sourceId }: { sourceId: string }) {
 }
 
 interface SkillTileProps {
-  skill: SkillSummary;
+  skill: WorkflowSummary;
   onUninstall: () => void;
 }
 
@@ -267,7 +267,7 @@ export default function SkillsExplorerTab({ onToast }: SkillsExplorerTabProps) {
   const { t } = useT();
   const [view, setView] = useState<ExplorerView>('registry');
 
-  const [skills, setSkills] = useState<SkillSummary[]>([]);
+  const [skills, setSkills] = useState<WorkflowSummary[]>([]);
   const [skillsLoading, setSkillsLoading] = useState(true);
   const [skillsError, setSkillsError] = useState<string | null>(null);
 
@@ -279,14 +279,14 @@ export default function SkillsExplorerTab({ onToast }: SkillsExplorerTabProps) {
   const [searchQuery, setSearchQuery] = useState('');
   const [formatFilter, setFormatFilter] = useState<string>('all');
   const [installDialogOpen, setInstallDialogOpen] = useState(false);
-  const [uninstallTarget, setUninstallTarget] = useState<SkillSummary | null>(null);
+  const [uninstallTarget, setUninstallTarget] = useState<WorkflowSummary | null>(null);
 
   const fetchSkills = useCallback(async () => {
     log('fetchSkills: start');
     setSkillsLoading(true);
     setSkillsError(null);
     try {
-      const result = await skillsApi.listSkills();
+      const result = await workflowsApi.listWorkflows();
       log('fetchSkills: count=%d', result.length);
       setSkills(result);
     } catch (err) {
@@ -368,16 +368,16 @@ export default function SkillsExplorerTab({ onToast }: SkillsExplorerTabProps) {
   }, [catalog]);
 
   const handleInstalled = useCallback(
-    (result: InstallSkillFromUrlResult) => {
-      log('handleInstalled: newSkills=%d', result.newSkills.length);
+    (result: InstallWorkflowFromUrlResult) => {
+      log('handleInstalled: newSkills=%d', result.newWorkflows.length);
       void fetchSkills();
-      if (result.newSkills.length > 0) {
+      if (result.newWorkflows.length > 0) {
         onToast?.({
           type: 'success',
           title: t('skills.install.installComplete'),
           message: t('skills.install.successDiscovered').replace(
             '{count}',
-            String(result.newSkills.length)
+            String(result.newWorkflows.length)
           ),
         });
       }

app/src/components/skills/UninstallSkillConfirmDialog.tsx

@@ -3,7 +3,7 @@
  * ---------------------------
  *
  * Small centered confirm modal for destructive uninstall of a user-scope
- * SKILL.md skill. Wraps `skillsApi.uninstallSkill` which calls
+ * SKILL.md skill. Wraps `workflowsApi.uninstallWorkflow` which calls
  * `openhuman.workflows_uninstall` on the Rust side — that RPC only accepts
  * user-scope installs (`~/.openhuman/skills/<name>/`) and refuses project
  * and legacy scopes. The card that opens this dialog is responsible for
@@ -29,23 +29,23 @@ import debug from 'debug';
 
 import { useT } from '../../lib/i18n/I18nContext';
 import {
-  skillsApi,
-  type SkillSummary,
-  type UninstallSkillResult,
-} from '../../services/api/skillsApi';
+  workflowsApi,
+  type WorkflowSummary,
+  type UninstallWorkflowResult,
+} from '../../services/api/workflowsApi';
 import { trackEvent } from '../../services/analytics';
 
 const log = debug('skills:uninstall-dialog');
 
 interface Props {
-  skill: SkillSummary;
+  skill: WorkflowSummary;
   onClose: () => void;
   /**
    * Fires when the backend reports the uninstall succeeded. Parent is
    * responsible for refetching the skills list and closing any detail
    * panels that were showing this skill.
    */
-  onUninstalled: (result: UninstallSkillResult) => void;
+  onUninstalled: (result: UninstallWorkflowResult) => void;
 }
 
 export default function UninstallSkillConfirmDialog({ skill, onClose, onUninstalled }: Props) {
@@ -82,7 +82,7 @@ export default function UninstallSkillConfirmDialog({ skill, onClose, onUninstal
       // `skill.id` is the on-disk slug (directory under ~/.openhuman/skills/).
       // `skill.name` is the frontmatter display name and may diverge from the
       // slug — the backend resolves by slug, so pass `id`.
-      const result = await skillsApi.uninstallSkill(skill.id);
+      const result = await workflowsApi.uninstallWorkflow(skill.id);
       log('confirm: done removedPath=%s', result.removedPath);
       trackEvent('skill_uninstall', { skill_id: skill.id });
       onUninstalled(result);

app/src/components/skills/WorkflowRunnerBody.tsx

@@ -21,11 +21,11 @@ import { SCHEDULE_PRESETS } from '../../lib/cron/schedulePresets';
 import {
   type RunLogSlice,
   type ScannedRun,
-  type SkillDescription,
-  type SkillRunStarted,
-  type SkillSummary,
-  skillsApi,
-} from '../../services/api/skillsApi';
+  type WorkflowDescription,
+  type WorkflowRunStarted,
+  type WorkflowSummary,
+  workflowsApi,
+} from '../../services/api/workflowsApi';
 import {
   type CoreCronJob,
   type CoreCronRun,
@@ -38,7 +38,7 @@ import {
 import CreateSkillModal from './CreateSkillModal';
 import BranchPicker from './inputs/BranchPicker';
 import RepoPicker from './inputs/RepoPicker';
-import { isGithubGateFailure, parseSkillRunError } from './preflightGate';
+import { isGithubGateFailure, parseWorkflowRunError } from './preflightGate';
 import ScheduledCronCard from './ScheduledCronCard';
 import SmartIssuePicker from './SmartIssuePicker';
 
@@ -95,7 +95,7 @@ type InputValue = string | number | boolean;
 interface RunState {
   status: 'idle' | 'submitting' | 'started' | 'error';
   message?: string;
-  result?: SkillRunStarted;
+  result?: WorkflowRunStarted;
 }
 
 
@@ -165,7 +165,7 @@ export function parseScheduledInputs(
 /**
  * Default form value for an input based on its declared type. Strings/
  * integers default to empty (renders as placeholder); booleans to false.
- * `runSkill` later trims and drops empty optional fields before sending
+ * `runWorkflow` later trims and drops empty optional fields before sending
  * them over the wire.
  */
 function defaultForType(type: string): InputValue {
@@ -175,13 +175,13 @@ function defaultForType(type: string): InputValue {
 }
 
 /**
- * Project the form-state map back into the JSON inputs shape `skills_run`
+ * Project the form-state map back into the JSON inputs shape `workflows_run`
  * expects: trim strings, coerce integer-typed fields to numbers, drop
  * empty optional fields entirely (so the backend sees them as "not
  * provided" rather than `""`).
  */
 function buildInputsPayload(
-  description: SkillDescription,
+  description: WorkflowDescription,
   values: Record<string, InputValue>
 ): Record<string, unknown> {
   const out: Record<string, unknown> = {};
@@ -225,7 +225,7 @@ export interface SkillsRunnerBodyProps {
    * the skill picker. Defaults to the Settings-panel description so
    * the original placement is unchanged. (Named `headerText` rather
    * than `description` to avoid shadowing the internal `description`
-   * state that holds the resolved `SkillDescription` for the picked
+   * state that holds the resolved `WorkflowDescription` for the picked
    * skill.)
    */
   headerText?: string;
@@ -241,7 +241,7 @@ export const WorkflowRunnerBody = ({ headerText, className }: SkillsRunnerBodyPr
   const { t } = useT();
 
   // Skill catalog (loaded once on mount)
-  const [skills, setSkills] = useState<SkillSummary[]>([]);
+  const [skills, setSkills] = useState<WorkflowSummary[]>([]);
   const [skillsLoading, setSkillsLoading] = useState(false);
   const [skillsError, setSkillsError] = useState<string | null>(null);
   // Edit-this-workflow modal (only meaningful when locked to a workflow).
@@ -263,7 +263,7 @@ export const WorkflowRunnerBody = ({ headerText, className }: SkillsRunnerBodyPr
     !!searchParams.get('workflow') &&
     (searchParams.get('lock') === '1' || searchParams.get('focus') === 'schedule');
   const selectedWorkflow = skills.find(s => s.id === selectedSkillId);
-  const [description, setDescription] = useState<SkillDescription | null>(null);
+  const [description, setDescription] = useState<WorkflowDescription | null>(null);
   const [descLoading, setDescLoading] = useState(false);
   const [descError, setDescError] = useState<string | null>(null);
 
@@ -449,13 +449,13 @@ export const WorkflowRunnerBody = ({ headerText, className }: SkillsRunnerBodyPr
     return () => clearTimeout(timer);
   }, [searchParams, selectedSkillId]);
 
-  // ── Initial load: skills_list ──────────────────────────────────────
+  // ── Initial load: workflows_list ──────────────────────────────────────
   useEffect(() => {
     let cancelled = false;
     setSkillsLoading(true);
     setSkillsError(null);
-    skillsApi
-      .listSkills()
+    workflowsApi
+      .listWorkflows()
       .then((list) => {
         if (cancelled) return;
         // Hide the codegraph-smoke skill — internal smoke-test only.
@@ -466,7 +466,7 @@ export const WorkflowRunnerBody = ({ headerText, className }: SkillsRunnerBodyPr
       .catch((err: unknown) => {
         if (cancelled) return;
         const msg = err instanceof Error ? err.message : String(err);
-        log('listSkills error: %s', msg);
+        log('listWorkflows error: %s', msg);
         setSkillsError(msg);
       })
       .finally(() => {
@@ -477,7 +477,7 @@ export const WorkflowRunnerBody = ({ headerText, className }: SkillsRunnerBodyPr
     };
   }, []);
 
-  // ── On selection: skills_describe ──────────────────────────────────
+  // ── On selection: workflows_describe ──────────────────────────────────
   useEffect(() => {
     if (!selectedSkillId) {
       setDescription(null);
@@ -488,8 +488,8 @@ export const WorkflowRunnerBody = ({ headerText, className }: SkillsRunnerBodyPr
     setDescLoading(true);
     setDescError(null);
     setRun({ status: 'idle' });
-    skillsApi
-      .describeSkill(selectedSkillId)
+    workflowsApi
+      .describeWorkflow(selectedSkillId)
       .then((desc) => {
         if (cancelled) return;
         setDescription(desc);
@@ -504,7 +504,7 @@ export const WorkflowRunnerBody = ({ headerText, className }: SkillsRunnerBodyPr
       .catch((err: unknown) => {
         if (cancelled) return;
         const msg = err instanceof Error ? err.message : String(err);
-        log('describeSkill error: %s', msg);
+        log('describeWorkflow error: %s', msg);
         setDescError(msg);
       })
       .finally(() => {
@@ -540,7 +540,7 @@ export const WorkflowRunnerBody = ({ headerText, className }: SkillsRunnerBodyPr
   const handleStopRun = useCallback(async (runId: string) => {
     log('stop run runId=%s', runId);
     try {
-      await skillsApi.cancelRun(runId);
+      await workflowsApi.cancelRun(runId);
     } catch (err) {
       log('cancelRun error: %s', err instanceof Error ? err.message : String(err));
     }
@@ -552,7 +552,7 @@ export const WorkflowRunnerBody = ({ headerText, className }: SkillsRunnerBodyPr
     // Re-entry guard: a second click before React applies the disabled state
     // would otherwise fire `workflows_run` twice and spawn two real runs.
     if (runSubmitGuardRef.current) {
-      log('runSkill: ignoring re-entrant click while a run is starting');
+      log('runWorkflow: ignoring re-entrant click while a run is starting');
       return;
     }
     if (missingRequired.length > 0) {
@@ -567,8 +567,8 @@ export const WorkflowRunnerBody = ({ headerText, className }: SkillsRunnerBodyPr
     setRun({ status: 'submitting' });
     try {
       const inputs = buildInputsPayload(description, formValues);
-      log('runSkill %s inputs=%o', description.id, inputs);
-      const result = await skillsApi.runSkill(description.id, inputs);
+      log('runWorkflow %s inputs=%o', description.id, inputs);
+      const result = await workflowsApi.runWorkflow(description.id, inputs);
       setRun({ status: 'started', result });
       // Surface the new run in "Recent runs" without a manual refresh, and
       // hold the guard through a short cooldown so a second click can't spawn
@@ -577,7 +577,7 @@ export const WorkflowRunnerBody = ({ headerText, className }: SkillsRunnerBodyPr
       releaseRunGuard(2500);
     } catch (err: unknown) {
       const msg = err instanceof Error ? err.message : String(err);
-      log('runSkill error: %s', msg);
+      log('runWorkflow error: %s', msg);
       setRun({ status: 'error', message: msg });
       releaseRunGuard(0); // allow immediate retry on failure
     }
@@ -587,7 +587,7 @@ export const WorkflowRunnerBody = ({ headerText, className }: SkillsRunnerBodyPr
   useEffect(() => {
     let cancelled = false;
     setRecentRunsLoading(true);
-    skillsApi
+    workflowsApi
       .recentRuns(selectedSkillId || undefined, 10)
       .then((list) => {
         if (cancelled) return;
@@ -701,7 +701,7 @@ export const WorkflowRunnerBody = ({ headerText, className }: SkillsRunnerBodyPr
             error: null,
           },
         }));
-        const slice: RunLogSlice = await skillsApi.readRunLog(runId, fromOffset);
+        const slice: RunLogSlice = await workflowsApi.readRunLog(runId, fromOffset);
         if (cancelled) return;
         setViewer((prev) => {
           const prior = prev[runId]?.content ?? '';
@@ -784,7 +784,7 @@ export const WorkflowRunnerBody = ({ headerText, className }: SkillsRunnerBodyPr
       );
       try {
         log('runJobNow: running %s directly with %o', selectedSkillId, inputs);
-        await skillsApi.runSkill(selectedSkillId, inputs);
+        await workflowsApi.runWorkflow(selectedSkillId, inputs);
         scheduleRecentRunsRefresh();
         releaseRunGuard(2500);
       } catch (err: unknown) {
@@ -908,7 +908,7 @@ export const WorkflowRunnerBody = ({ headerText, className }: SkillsRunnerBodyPr
   // instead of a plain text input. Falls through to the type-based
   // string/integer/boolean handling for everything else.
   const renderField = (
-    inp: SkillDescription['inputs'][number],
+    inp: WorkflowDescription['inputs'][number],
     value: InputValue,
     onChange: (next: InputValue) => void
   ) => {
@@ -1065,7 +1065,7 @@ export const WorkflowRunnerBody = ({ headerText, className }: SkillsRunnerBodyPr
           )}
           {skillsError && (
             <p className="text-xs text-red-600 dark:text-red-400 mt-1">
-              {t('settings.skillsRunner.error.listSkills')} {skillsError}
+              {t('settings.skillsRunner.error.listWorkflows')} {skillsError}
             </p>
           )}
         </div>
@@ -1199,7 +1199,7 @@ export const WorkflowRunnerBody = ({ headerText, className }: SkillsRunnerBodyPr
                     // failed" pill above the body so the user knows
                     // this isn't a generic crash — there's a concrete
                     // remediation the body describes.
-                    const parsed = parseSkillRunError(run.message);
+                    const parsed = parseWorkflowRunError(run.message);
                     const isGateFailure = isGithubGateFailure(parsed);
                     return (
                       <div
@@ -1588,12 +1588,12 @@ export const WorkflowRunnerBody = ({ headerText, className }: SkillsRunnerBodyPr
           onClose={() => setEditOpen(false)}
           onCreated={() => {
             setEditOpen(false);
-            void skillsApi
-              .listSkills()
+            void workflowsApi
+              .listWorkflows()
               .then((list) => setSkills(list.filter((s) => s.id !== 'codegraph-smoke')))
               .catch(() => {});
-            void skillsApi
-              .describeSkill(selectedSkillId)
+            void workflowsApi
+              .describeWorkflow(selectedSkillId)
               .then(setDescription)
               .catch(() => {});
           }}

app/src/components/skills/__tests__/CreateSkillModal.test.tsx

@@ -6,7 +6,7 @@
  * - Escape key closes (but not while submitting).
  * - Backdrop click closes (but not while submitting).
  * - Submit is disabled when name or description is empty.
- * - Submit rekeys `allowedTools` → `'allowed-tools'` via skillsApi.createSkill.
+ * - Submit rekeys `allowedTools` → `'allowed-tools'` via workflowsApi.createWorkflow.
  * - Submit calls `onCreated` with the returned skill.
  * - Submit failure surfaces an error banner and re-enables the button.
  * - Slug preview updates as the name changes.
@@ -14,18 +14,18 @@
 import { act, fireEvent, render, screen, waitFor } from '@testing-library/react';
 import { beforeEach, describe, expect, it, vi } from 'vitest';
 
-import type { SkillSummary } from '../../../services/api/skillsApi';
+import type { WorkflowSummary } from '../../../services/api/workflowsApi';
 import CreateSkillModal from '../CreateSkillModal';
 
-vi.mock('../../../services/api/skillsApi', () => ({
-  skillsApi: {
-    createSkill: vi.fn(),
-    updateSkill: vi.fn(),
-    describeSkill: vi.fn().mockResolvedValue({ id: 'e', name: 'e', when_to_use: '', inputs: [] }),
+vi.mock('../../../services/api/workflowsApi', () => ({
+  workflowsApi: {
+    createWorkflow: vi.fn(),
+    updateWorkflow: vi.fn(),
+    describeWorkflow: vi.fn().mockResolvedValue({ id: 'e', name: 'e', when_to_use: '', inputs: [] }),
   },
 }));
 
-function builtSkill(overrides: Partial<SkillSummary> = {}): SkillSummary {
+function builtSkill(overrides: Partial<WorkflowSummary> = {}): WorkflowSummary {
   return {
     id: 'my-skill',
     name: 'My Skill',
@@ -49,8 +49,8 @@ function builtSkill(overrides: Partial<SkillSummary> = {}): SkillSummary {
 
 describe('CreateSkillModal', () => {
   beforeEach(async () => {
-    const { skillsApi } = await import('../../../services/api/skillsApi');
-    vi.mocked(skillsApi.createSkill).mockReset();
+    const { workflowsApi } = await import('../../../services/api/workflowsApi');
+    vi.mocked(workflowsApi.createWorkflow).mockReset();
   });
 
   it('renders title and required fields', () => {
@@ -94,13 +94,13 @@ describe('CreateSkillModal', () => {
     // form is now Name + Description + the `[[inputs]]` editor only — see
     // ScheduledCronCard / CreateWorkflowForm.tsx), so the inputs are no longer
     // collectable from the modal UI. The rekey itself still happens in
-    // `skillsApi.createSkill` (services/api/skillsApi.ts → params build) and
-    // is covered by the skillsApi unit tests; this test now just guards the
-    // modal's submit-pipeline shape: name + description → createSkill →
+    // `workflowsApi.createWorkflow` (services/api/workflowsApi.ts → params build) and
+    // is covered by the workflowsApi unit tests; this test now just guards the
+    // modal's submit-pipeline shape: name + description → createWorkflow →
     // onCreated.
-    const { skillsApi } = await import('../../../services/api/skillsApi');
+    const { workflowsApi } = await import('../../../services/api/workflowsApi');
     const created = builtSkill();
-    vi.mocked(skillsApi.createSkill).mockResolvedValueOnce(created);
+    vi.mocked(workflowsApi.createWorkflow).mockResolvedValueOnce(created);
 
     const onCreated = vi.fn();
     const onClose = vi.fn();
@@ -114,7 +114,7 @@ describe('CreateSkillModal', () => {
       fireEvent.click(submit);
     });
 
-    expect(vi.mocked(skillsApi.createSkill)).toHaveBeenCalledWith(
+    expect(vi.mocked(workflowsApi.createWorkflow)).toHaveBeenCalledWith(
       expect.objectContaining({
         name: 'My Skill',
         description: 'does stuff',
@@ -125,8 +125,8 @@ describe('CreateSkillModal', () => {
   });
 
   it('surfaces error and re-enables submit on failure', async () => {
-    const { skillsApi } = await import('../../../services/api/skillsApi');
-    vi.mocked(skillsApi.createSkill).mockRejectedValueOnce(new Error('slug already exists'));
+    const { workflowsApi } = await import('../../../services/api/workflowsApi');
+    vi.mocked(workflowsApi.createWorkflow).mockRejectedValueOnce(new Error('slug already exists'));
 
     render(<CreateSkillModal onClose={vi.fn()} onCreated={vi.fn()} />);
     fireEvent.change(screen.getByLabelText(/Name/), { target: { value: 'dup' } });

app/src/components/skills/__tests__/CreateWorkflowForm.test.tsx

@@ -6,7 +6,7 @@
  * and the /skills/new page can rely on it.
  *
  * Covers:
- *  - submit calls skillsApi.createSkill with the trimmed/normalised
+ *  - submit calls workflowsApi.createWorkflow with the trimmed/normalised
  *    payload (CSVs split, optional fields omitted when empty).
  *  - onStateChange is called with validity + submitting flags so
  *    wrappers can sync their submit button's disabled state.
@@ -20,16 +20,16 @@ const stableT = (key: string) => key;
 vi.mock('../../../lib/i18n/I18nContext', () => ({ useT: () => ({ t: stableT }) }));
 
 const hoisted = vi.hoisted(() => ({
-  createSkill: vi.fn(),
-  updateSkill: vi.fn(),
-  describeSkill: vi.fn(),
+  createWorkflow: vi.fn(),
+  updateWorkflow: vi.fn(),
+  describeWorkflow: vi.fn(),
 }));
 
-vi.mock('../../../services/api/skillsApi', () => ({
-  skillsApi: {
-    createSkill: hoisted.createSkill,
-    updateSkill: hoisted.updateSkill,
-    describeSkill: hoisted.describeSkill,
+vi.mock('../../../services/api/workflowsApi', () => ({
+  workflowsApi: {
+    createWorkflow: hoisted.createWorkflow,
+    updateWorkflow: hoisted.updateWorkflow,
+    describeWorkflow: hoisted.describeWorkflow,
   },
 }));
 
@@ -57,7 +57,7 @@ describe('previewSlug', () => {
 
 describe('CreateWorkflowForm', () => {
   beforeEach(() => {
-    hoisted.createSkill.mockReset();
+    hoisted.createWorkflow.mockReset();
   });
 
   it('renders required fields and the slug preview updates as the name changes', () => {
@@ -98,7 +98,7 @@ describe('CreateWorkflowForm', () => {
     // fields were dropped. Anyone needing project-scoped or
     // tagged skills edits the workspace SKILL.md directly.
     const created = { id: 'my-skill', name: 'My Skill', scope: 'user', legacy: false };
-    hoisted.createSkill.mockResolvedValue(created);
+    hoisted.createWorkflow.mockResolvedValue(created);
     const onCreated = vi.fn();
 
     render(<CreateWorkflowForm formId={FORM_ID} onCreated={onCreated} />);
@@ -116,7 +116,7 @@ describe('CreateWorkflowForm', () => {
     fireEvent.submit(document.getElementById(FORM_ID)!);
 
     await waitFor(() => {
-      expect(hoisted.createSkill).toHaveBeenCalledWith({
+      expect(hoisted.createWorkflow).toHaveBeenCalledWith({
         name: 'My Skill',
         description: 'Does the thing.',
         scope: 'user',
@@ -126,7 +126,7 @@ describe('CreateWorkflowForm', () => {
   });
 
   it('includes whenToUse in the payload when the trigger field is filled', async () => {
-    hoisted.createSkill.mockResolvedValue({ id: 'wf', name: 'wf', scope: 'user', legacy: false });
+    hoisted.createWorkflow.mockResolvedValue({ id: 'wf', name: 'wf', scope: 'user', legacy: false });
     render(<CreateWorkflowForm formId={FORM_ID} onCreated={vi.fn()} />);
 
     fireEvent.change(screen.getByLabelText(/skills.create.name/i), {
@@ -141,7 +141,7 @@ describe('CreateWorkflowForm', () => {
     fireEvent.submit(document.getElementById(FORM_ID)!);
 
     await waitFor(() => {
-      expect(hoisted.createSkill).toHaveBeenCalledWith({
+      expect(hoisted.createWorkflow).toHaveBeenCalledWith({
         name: 'Triage',
         description: 'Summarise the inbox.',
         scope: 'user',
@@ -150,8 +150,8 @@ describe('CreateWorkflowForm', () => {
     });
   });
 
-  it('surfaces the Rust error message in an alert when createSkill rejects', async () => {
-    hoisted.createSkill.mockRejectedValue(new Error('slug already exists'));
+  it('surfaces the Rust error message in an alert when createWorkflow rejects', async () => {
+    hoisted.createWorkflow.mockRejectedValue(new Error('slug already exists'));
     render(<CreateWorkflowForm formId={FORM_ID} onCreated={vi.fn()} />);
 
     fireEvent.change(screen.getByLabelText(/skills.create.name/i), {
@@ -166,18 +166,18 @@ describe('CreateWorkflowForm', () => {
     expect(alert).toHaveTextContent('slug already exists');
   });
 
-  it('does not call createSkill if the form is invalid (no name)', async () => {
+  it('does not call createWorkflow if the form is invalid (no name)', async () => {
     render(<CreateWorkflowForm formId={FORM_ID} onCreated={vi.fn()} />);
     fireEvent.submit(document.getElementById(FORM_ID)!);
     // Give the microtask queue a tick — should still be 0.
     await Promise.resolve();
-    expect(hoisted.createSkill).not.toHaveBeenCalled();
+    expect(hoisted.createWorkflow).not.toHaveBeenCalled();
   });
 
   // ── Inputs editor ───────────────────────────────────────────────────
   // The form gained an optional [[inputs]] editor in 5d77839f. These
   // tests pin its contract end-to-end: the rows the user adds become the
-  // `inputs` field on the createSkill payload, name validation blocks
+  // `inputs` field on the createWorkflow payload, name validation blocks
   // submission, and removing a row drops it from the payload.
 
   /** Fill name + description so the rest of the form is submittable. */
@@ -199,23 +199,23 @@ describe('CreateWorkflowForm', () => {
   }
 
   it('zero inputs submits a payload without an `inputs` field', async () => {
-    hoisted.createSkill.mockResolvedValue({ id: 'x', name: 'x', scope: 'user', legacy: false });
+    hoisted.createWorkflow.mockResolvedValue({ id: 'x', name: 'x', scope: 'user', legacy: false });
     render(<CreateWorkflowForm formId={FORM_ID} onCreated={vi.fn()} />);
     fillRequiredFields();
     fireEvent.submit(document.getElementById(FORM_ID)!);
     await waitFor(() => {
-      expect(hoisted.createSkill).toHaveBeenCalledWith({
+      expect(hoisted.createWorkflow).toHaveBeenCalledWith({
         name: 'My Skill',
         description: 'Does the thing.',
         scope: 'user',
       });
     });
-    const payload = hoisted.createSkill.mock.calls[0]![0] as Record<string, unknown>;
+    const payload = hoisted.createWorkflow.mock.calls[0]![0] as Record<string, unknown>;
     expect(payload).not.toHaveProperty('inputs');
   });
 
   it('one filled input row ships in the payload — additional inputs default to required: false', async () => {
-    hoisted.createSkill.mockResolvedValue({ id: 'x', name: 'x', scope: 'user', legacy: false });
+    hoisted.createWorkflow.mockResolvedValue({ id: 'x', name: 'x', scope: 'user', legacy: false });
     render(<CreateWorkflowForm formId={FORM_ID} onCreated={vi.fn()} />);
     fillRequiredFields();
 
@@ -227,7 +227,7 @@ describe('CreateWorkflowForm', () => {
 
     fireEvent.submit(document.getElementById(FORM_ID)!);
     await waitFor(() => {
-      expect(hoisted.createSkill).toHaveBeenCalledWith({
+      expect(hoisted.createWorkflow).toHaveBeenCalledWith({
         name: 'My Skill',
         description: 'Does the thing.',
         scope: 'user',
@@ -238,7 +238,7 @@ describe('CreateWorkflowForm', () => {
   });
 
   it('blocks submission when an added input row has no description', async () => {
-    hoisted.createSkill.mockResolvedValue({ id: 'x', name: 'x', scope: 'user', legacy: false });
+    hoisted.createWorkflow.mockResolvedValue({ id: 'x', name: 'x', scope: 'user', legacy: false });
     render(<CreateWorkflowForm formId={FORM_ID} onCreated={vi.fn()} />);
     fillRequiredFields();
 
@@ -249,7 +249,7 @@ describe('CreateWorkflowForm', () => {
     fireEvent.change(nameInput, { target: { value: 'repo' } });
     fireEvent.submit(document.getElementById(FORM_ID)!);
     await Promise.resolve();
-    expect(hoisted.createSkill).not.toHaveBeenCalled();
+    expect(hoisted.createWorkflow).not.toHaveBeenCalled();
     expect(screen.getByText(/descriptionError/i)).toBeInTheDocument();
 
     // Fill the description → now it submits.
@@ -257,12 +257,12 @@ describe('CreateWorkflowForm', () => {
     fireEvent.change(descInput, { target: { value: 'owner/name' } });
     fireEvent.submit(document.getElementById(FORM_ID)!);
     await waitFor(() => {
-      expect(hoisted.createSkill).toHaveBeenCalledTimes(1);
+      expect(hoisted.createWorkflow).toHaveBeenCalledTimes(1);
     });
   });
 
   it('ticking Required flips the row to required: true', async () => {
-    hoisted.createSkill.mockResolvedValue({ id: 'x', name: 'x', scope: 'user', legacy: false });
+    hoisted.createWorkflow.mockResolvedValue({ id: 'x', name: 'x', scope: 'user', legacy: false });
     render(<CreateWorkflowForm formId={FORM_ID} onCreated={vi.fn()} />);
     fillRequiredFields();
 
@@ -277,7 +277,7 @@ describe('CreateWorkflowForm', () => {
 
     fireEvent.submit(document.getElementById(FORM_ID)!);
     await waitFor(() => {
-      expect(hoisted.createSkill).toHaveBeenCalledWith(
+      expect(hoisted.createWorkflow).toHaveBeenCalledWith(
         expect.objectContaining({
           inputs: [{ name: 'repo', required: true, description: 'owner/name' }],
         })
@@ -286,7 +286,7 @@ describe('CreateWorkflowForm', () => {
   });
 
   it('blocks submission while any row has an invalid name', async () => {
-    hoisted.createSkill.mockResolvedValue({ id: 'x', name: 'x', scope: 'user', legacy: false });
+    hoisted.createWorkflow.mockResolvedValue({ id: 'x', name: 'x', scope: 'user', legacy: false });
     render(<CreateWorkflowForm formId={FORM_ID} onCreated={vi.fn()} />);
     fillRequiredFields();
 
@@ -294,7 +294,7 @@ describe('CreateWorkflowForm', () => {
     // Add a row but leave the name empty — submission must be blocked.
     fireEvent.submit(document.getElementById(FORM_ID)!);
     await Promise.resolve();
-    expect(hoisted.createSkill).not.toHaveBeenCalled();
+    expect(hoisted.createWorkflow).not.toHaveBeenCalled();
 
     // Fill the name with an invalid character (leading digit).
     const row = lastRow();
@@ -302,14 +302,14 @@ describe('CreateWorkflowForm', () => {
     fireEvent.change(nameInput, { target: { value: '2repo' } });
     fireEvent.submit(document.getElementById(FORM_ID)!);
     await Promise.resolve();
-    expect(hoisted.createSkill).not.toHaveBeenCalled();
+    expect(hoisted.createWorkflow).not.toHaveBeenCalled();
 
     // Inline error visible.
     expect(screen.getByText(/nameError/i)).toBeInTheDocument();
   });
 
   it('remove row drops it from the payload — submission then succeeds', async () => {
-    hoisted.createSkill.mockResolvedValue({ id: 'x', name: 'x', scope: 'user', legacy: false });
+    hoisted.createWorkflow.mockResolvedValue({ id: 'x', name: 'x', scope: 'user', legacy: false });
     render(<CreateWorkflowForm formId={FORM_ID} onCreated={vi.fn()} />);
     fillRequiredFields();
 
@@ -324,14 +324,14 @@ describe('CreateWorkflowForm', () => {
     // `inputs` field at all (back to the no-inputs payload shape).
     fireEvent.submit(document.getElementById(FORM_ID)!);
     await waitFor(() => {
-      expect(hoisted.createSkill).toHaveBeenCalledTimes(1);
+      expect(hoisted.createWorkflow).toHaveBeenCalledTimes(1);
     });
-    const payload = hoisted.createSkill.mock.calls[0]![0] as Record<string, unknown>;
+    const payload = hoisted.createWorkflow.mock.calls[0]![0] as Record<string, unknown>;
     expect(payload).not.toHaveProperty('inputs');
   });
 
   it('integer + required=false (the default) carry through the type + required flags', async () => {
-    hoisted.createSkill.mockResolvedValue({ id: 'x', name: 'x', scope: 'user', legacy: false });
+    hoisted.createWorkflow.mockResolvedValue({ id: 'x', name: 'x', scope: 'user', legacy: false });
     render(<CreateWorkflowForm formId={FORM_ID} onCreated={vi.fn()} />);
     fillRequiredFields();
 
@@ -347,7 +347,7 @@ describe('CreateWorkflowForm', () => {
 
     fireEvent.submit(document.getElementById(FORM_ID)!);
     await waitFor(() => {
-      expect(hoisted.createSkill).toHaveBeenCalledWith(
+      expect(hoisted.createWorkflow).toHaveBeenCalledWith(
         expect.objectContaining({
           inputs: [
             { name: 'issue', required: false, type: 'integer', description: 'Issue number to work on' },
@@ -360,9 +360,9 @@ describe('CreateWorkflowForm', () => {
 
 describe('CreateWorkflowForm — edit mode', () => {
   beforeEach(() => {
-    hoisted.createSkill.mockReset();
-    hoisted.updateSkill.mockReset();
-    hoisted.describeSkill.mockReset();
+    hoisted.createWorkflow.mockReset();
+    hoisted.updateWorkflow.mockReset();
+    hoisted.describeWorkflow.mockReset();
   });
 
   const editingSummary = {
@@ -384,14 +384,14 @@ describe('CreateWorkflowForm — edit mode', () => {
     warnings: [],
   };
 
-  it('prefills from the summary + describe, then submits via updateSkill', async () => {
-    hoisted.describeSkill.mockResolvedValue({
+  it('prefills from the summary + describe, then submits via updateWorkflow', async () => {
+    hoisted.describeWorkflow.mockResolvedValue({
       id: 'wf-edit',
       name: 'wf-edit',
       when_to_use: 'edit trigger',
       inputs: [{ name: 'repo', type: 'string', required: true, description: 'r' }],
     });
-    hoisted.updateSkill.mockResolvedValue({
+    hoisted.updateWorkflow.mockResolvedValue({
       id: 'wf-edit',
       name: 'wf-edit',
       scope: 'user',
@@ -402,13 +402,13 @@ describe('CreateWorkflowForm — edit mode', () => {
     render(<CreateWorkflowForm formId={FORM_ID} editing={editingSummary} onCreated={onCreated} />);
 
     // Prefill: describe is fetched and name is populated from the summary.
-    await waitFor(() => expect(hoisted.describeSkill).toHaveBeenCalledWith('wf-edit'));
+    await waitFor(() => expect(hoisted.describeWorkflow).toHaveBeenCalledWith('wf-edit'));
     const nameInput = screen.getByLabelText(/skills.create.name/i) as HTMLInputElement;
     await waitFor(() => expect(nameInput.value).toBe('wf-edit'));
 
     fireEvent.submit(document.getElementById(FORM_ID)!);
     await waitFor(() =>
-      expect(hoisted.updateSkill).toHaveBeenCalledWith(
+      expect(hoisted.updateWorkflow).toHaveBeenCalledWith(
         expect.objectContaining({
           name: 'wf-edit',
           whenToUse: 'edit trigger',
@@ -418,7 +418,7 @@ describe('CreateWorkflowForm — edit mode', () => {
         })
       )
     );
-    expect(hoisted.createSkill).not.toHaveBeenCalled();
+    expect(hoisted.createWorkflow).not.toHaveBeenCalled();
     await waitFor(() => expect(onCreated).toHaveBeenCalled());
   });
 });

app/src/components/skills/__tests__/InstallSkillDialog.test.tsx

@@ -6,8 +6,8 @@
  * - Submit disabled until a well-formed https URL is entered.
  * - Shows inline error for non-https URLs.
  * - Rejects timeout outside 1–600.
- * - Submit forwards timeoutSecs to skillsApi.installSkillFromUrl.
- * - Success panel renders newSkills list + calls onInstalled.
+ * - Submit forwards timeoutSecs to workflowsApi.installWorkflowFromUrl.
+ * - Success panel renders newWorkflows list + calls onInstalled.
  * - Error panel categorizes known prefixes and shows the raw error in
  *   a details expander; unknown errors fall back to a generic title.
  */
@@ -16,16 +16,16 @@ import { beforeEach, describe, expect, it, vi } from 'vitest';
 
 import InstallSkillDialog from '../InstallSkillDialog';
 
-vi.mock('../../../services/api/skillsApi', () => ({
-  skillsApi: {
-    installSkillFromUrl: vi.fn(),
+vi.mock('../../../services/api/workflowsApi', () => ({
+  workflowsApi: {
+    installWorkflowFromUrl: vi.fn(),
   },
 }));
 
 describe('InstallSkillDialog', () => {
   beforeEach(async () => {
-    const { skillsApi } = await import('../../../services/api/skillsApi');
-    vi.mocked(skillsApi.installSkillFromUrl).mockReset();
+    const { workflowsApi } = await import('../../../services/api/workflowsApi');
+    vi.mocked(workflowsApi.installWorkflowFromUrl).mockReset();
   });
 
   it('renders title and URL input', () => {
@@ -72,13 +72,13 @@ describe('InstallSkillDialog', () => {
     expect(submit.disabled).toBe(false);
   });
 
-  it('forwards timeoutSecs to skillsApi and fires onInstalled on success', async () => {
-    const { skillsApi } = await import('../../../services/api/skillsApi');
-    vi.mocked(skillsApi.installSkillFromUrl).mockResolvedValueOnce({
+  it('forwards timeoutSecs to workflowsApi and fires onInstalled on success', async () => {
+    const { workflowsApi } = await import('../../../services/api/workflowsApi');
+    vi.mocked(workflowsApi.installWorkflowFromUrl).mockResolvedValueOnce({
       url: 'https://raw.githubusercontent.com/owner/repo/main/SKILL.md',
       stdout: 'added my-skill',
       stderr: '',
-      newSkills: ['my-skill'],
+      newWorkflows: ['my-skill'],
     });
 
     const onInstalled = vi.fn();
@@ -93,7 +93,7 @@ describe('InstallSkillDialog', () => {
       fireEvent.click(screen.getByRole('button', { name: /Install/ }));
     });
 
-    expect(vi.mocked(skillsApi.installSkillFromUrl)).toHaveBeenCalledWith({
+    expect(vi.mocked(workflowsApi.installWorkflowFromUrl)).toHaveBeenCalledWith({
       url: 'https://raw.githubusercontent.com/owner/repo/main/SKILL.md',
       timeoutSecs: 120,
     });
@@ -102,17 +102,17 @@ describe('InstallSkillDialog', () => {
     });
     expect(screen.getByText('my-skill')).toBeInTheDocument();
     expect(onInstalled).toHaveBeenCalledWith(
-      expect.objectContaining({ newSkills: ['my-skill'] })
+      expect.objectContaining({ newWorkflows: ['my-skill'] })
     );
   });
 
   it('omits timeoutSecs when field is blank', async () => {
-    const { skillsApi } = await import('../../../services/api/skillsApi');
-    vi.mocked(skillsApi.installSkillFromUrl).mockResolvedValueOnce({
+    const { workflowsApi } = await import('../../../services/api/workflowsApi');
+    vi.mocked(workflowsApi.installWorkflowFromUrl).mockResolvedValueOnce({
       url: 'https://raw.githubusercontent.com/owner/repo/main/SKILL.md',
       stdout: '',
       stderr: '',
-      newSkills: [],
+      newWorkflows: [],
     });
 
     render(<InstallSkillDialog onClose={vi.fn()} onInstalled={vi.fn()} />);
@@ -124,14 +124,14 @@ describe('InstallSkillDialog', () => {
       fireEvent.click(screen.getByRole('button', { name: /Install/ }));
     });
 
-    expect(vi.mocked(skillsApi.installSkillFromUrl)).toHaveBeenCalledWith({
+    expect(vi.mocked(workflowsApi.installWorkflowFromUrl)).toHaveBeenCalledWith({
       url: 'https://raw.githubusercontent.com/owner/repo/main/SKILL.md',
     });
   });
 
   it('shows generic title with raw error text on unknown error and re-enables submit', async () => {
-    const { skillsApi } = await import('../../../services/api/skillsApi');
-    vi.mocked(skillsApi.installSkillFromUrl).mockRejectedValueOnce(
+    const { workflowsApi } = await import('../../../services/api/workflowsApi');
+    vi.mocked(workflowsApi.installWorkflowFromUrl).mockRejectedValueOnce(
       new Error('unexpected: something weird happened')
     );
 
@@ -155,8 +155,8 @@ describe('InstallSkillDialog', () => {
   });
 
   it('categorizes "invalid SKILL.md:" errors with a friendly title and hint', async () => {
-    const { skillsApi } = await import('../../../services/api/skillsApi');
-    vi.mocked(skillsApi.installSkillFromUrl).mockRejectedValueOnce(
+    const { workflowsApi } = await import('../../../services/api/workflowsApi');
+    vi.mocked(workflowsApi.installWorkflowFromUrl).mockRejectedValueOnce(
       new Error('invalid SKILL.md: missing required field `description`')
     );
 
@@ -177,8 +177,8 @@ describe('InstallSkillDialog', () => {
   });
 
   it('categorizes "unsupported url form:" errors', async () => {
-    const { skillsApi } = await import('../../../services/api/skillsApi');
-    vi.mocked(skillsApi.installSkillFromUrl).mockRejectedValueOnce(
+    const { workflowsApi } = await import('../../../services/api/workflowsApi');
+    vi.mocked(workflowsApi.installWorkflowFromUrl).mockRejectedValueOnce(
       new Error('unsupported url form: path must end in .md, got "https://example.com/foo"')
     );
 

app/src/components/skills/__tests__/SkillsExplorerTab.test.tsx

@@ -2,14 +2,14 @@ import { act, fireEvent, render, screen, waitFor } from '@testing-library/react'
 import { beforeEach, describe, expect, it, vi } from 'vitest';
 
 import type { CatalogEntry } from '../../../services/api/skillRegistryApi';
-import type { SkillSummary } from '../../../services/api/skillsApi';
+import type { WorkflowSummary } from '../../../services/api/workflowsApi';
 import SkillsExplorerTab from '../SkillsExplorerTab';
 
-vi.mock('../../../services/api/skillsApi', () => ({
-  skillsApi: {
-    listSkills: vi.fn(),
-    installSkillFromUrl: vi.fn(),
-    uninstallSkill: vi.fn(),
+vi.mock('../../../services/api/workflowsApi', () => ({
+  workflowsApi: {
+    listWorkflows: vi.fn(),
+    installWorkflowFromUrl: vi.fn(),
+    uninstallWorkflow: vi.fn(),
   },
 }));
 
@@ -22,7 +22,7 @@ vi.mock('../../../services/api/skillRegistryApi', () => ({
   },
 }));
 
-const MOCK_SKILL: SkillSummary = {
+const MOCK_SKILL: WorkflowSummary = {
   id: 'test-skill',
   name: 'Test Skill',
   description: 'A test skill for unit testing',
@@ -41,7 +41,7 @@ const MOCK_SKILL: SkillSummary = {
   warnings: [],
 };
 
-const MOCK_PROJECT_SKILL: SkillSummary = {
+const MOCK_PROJECT_SKILL: WorkflowSummary = {
   ...MOCK_SKILL,
   id: 'project-skill',
   name: 'Project Skill',
@@ -72,23 +72,23 @@ async function switchToInstalled() {
 
 describe('SkillsExplorerTab', () => {
   beforeEach(async () => {
-    const { skillsApi } = await import('../../../services/api/skillsApi');
+    const { workflowsApi } = await import('../../../services/api/workflowsApi');
     const { skillRegistryApi } = await import(
       '../../../services/api/skillRegistryApi'
     );
-    vi.mocked(skillsApi.listSkills).mockReset();
-    vi.mocked(skillsApi.uninstallSkill).mockReset();
+    vi.mocked(workflowsApi.listWorkflows).mockReset();
+    vi.mocked(workflowsApi.uninstallWorkflow).mockReset();
     vi.mocked(skillRegistryApi.browse).mockReset();
     vi.mocked(skillRegistryApi.install).mockReset();
     vi.mocked(skillRegistryApi.browse).mockResolvedValue([]);
   });
 
   it('defaults to registry view and shows catalog entries', async () => {
-    const { skillsApi } = await import('../../../services/api/skillsApi');
+    const { workflowsApi } = await import('../../../services/api/workflowsApi');
     const { skillRegistryApi } = await import(
       '../../../services/api/skillRegistryApi'
     );
-    vi.mocked(skillsApi.listSkills).mockResolvedValue([]);
+    vi.mocked(workflowsApi.listWorkflows).mockResolvedValue([]);
     vi.mocked(skillRegistryApi.browse).mockResolvedValue([MOCK_CATALOG_ENTRY]);
 
     render(<SkillsExplorerTab />);
@@ -101,8 +101,8 @@ describe('SkillsExplorerTab', () => {
   });
 
   it('shows installed skills when switching to installed tab', async () => {
-    const { skillsApi } = await import('../../../services/api/skillsApi');
-    vi.mocked(skillsApi.listSkills).mockResolvedValue([MOCK_SKILL, MOCK_PROJECT_SKILL]);
+    const { workflowsApi } = await import('../../../services/api/workflowsApi');
+    vi.mocked(workflowsApi.listWorkflows).mockResolvedValue([MOCK_SKILL, MOCK_PROJECT_SKILL]);
 
     render(<SkillsExplorerTab />);
 
@@ -119,8 +119,8 @@ describe('SkillsExplorerTab', () => {
   });
 
   it('shows empty state when no installed skills', async () => {
-    const { skillsApi } = await import('../../../services/api/skillsApi');
-    vi.mocked(skillsApi.listSkills).mockResolvedValue([]);
+    const { workflowsApi } = await import('../../../services/api/workflowsApi');
+    vi.mocked(workflowsApi.listWorkflows).mockResolvedValue([]);
 
     render(<SkillsExplorerTab />);
     await switchToInstalled();
@@ -134,8 +134,8 @@ describe('SkillsExplorerTab', () => {
     const { skillRegistryApi } = await import(
       '../../../services/api/skillRegistryApi'
     );
-    const { skillsApi } = await import('../../../services/api/skillsApi');
-    vi.mocked(skillsApi.listSkills).mockResolvedValue([]);
+    const { workflowsApi } = await import('../../../services/api/workflowsApi');
+    vi.mocked(workflowsApi.listWorkflows).mockResolvedValue([]);
     vi.mocked(skillRegistryApi.browse).mockRejectedValue(new Error('Network error'));
 
     render(<SkillsExplorerTab />);
@@ -147,8 +147,8 @@ describe('SkillsExplorerTab', () => {
   });
 
   it('filters installed skills by search query', async () => {
-    const { skillsApi } = await import('../../../services/api/skillsApi');
-    vi.mocked(skillsApi.listSkills).mockResolvedValue([MOCK_SKILL, MOCK_PROJECT_SKILL]);
+    const { workflowsApi } = await import('../../../services/api/workflowsApi');
+    vi.mocked(workflowsApi.listWorkflows).mockResolvedValue([MOCK_SKILL, MOCK_PROJECT_SKILL]);
 
     render(<SkillsExplorerTab />);
     await switchToInstalled();
@@ -165,8 +165,8 @@ describe('SkillsExplorerTab', () => {
   });
 
   it('shows install from URL button', async () => {
-    const { skillsApi } = await import('../../../services/api/skillsApi');
-    vi.mocked(skillsApi.listSkills).mockResolvedValue([]);
+    const { workflowsApi } = await import('../../../services/api/workflowsApi');
+    vi.mocked(workflowsApi.listWorkflows).mockResolvedValue([]);
 
     render(<SkillsExplorerTab />);
 
@@ -176,8 +176,8 @@ describe('SkillsExplorerTab', () => {
   });
 
   it('shows uninstall button only for user-scope skills', async () => {
-    const { skillsApi } = await import('../../../services/api/skillsApi');
-    vi.mocked(skillsApi.listSkills).mockResolvedValue([MOCK_SKILL, MOCK_PROJECT_SKILL]);
+    const { workflowsApi } = await import('../../../services/api/workflowsApi');
+    vi.mocked(workflowsApi.listWorkflows).mockResolvedValue([MOCK_SKILL, MOCK_PROJECT_SKILL]);
 
     render(<SkillsExplorerTab />);
     await switchToInstalled();
@@ -191,8 +191,8 @@ describe('SkillsExplorerTab', () => {
   });
 
   it('displays version and tags in installed view', async () => {
-    const { skillsApi } = await import('../../../services/api/skillsApi');
-    vi.mocked(skillsApi.listSkills).mockResolvedValue([MOCK_SKILL]);
+    const { workflowsApi } = await import('../../../services/api/workflowsApi');
+    vi.mocked(workflowsApi.listWorkflows).mockResolvedValue([MOCK_SKILL]);
 
     render(<SkillsExplorerTab />);
     await switchToInstalled();
@@ -205,8 +205,8 @@ describe('SkillsExplorerTab', () => {
   });
 
   it('displays scope badges', async () => {
-    const { skillsApi } = await import('../../../services/api/skillsApi');
-    vi.mocked(skillsApi.listSkills).mockResolvedValue([MOCK_SKILL, MOCK_PROJECT_SKILL]);
+    const { workflowsApi } = await import('../../../services/api/workflowsApi');
+    vi.mocked(workflowsApi.listWorkflows).mockResolvedValue([MOCK_SKILL, MOCK_PROJECT_SKILL]);
 
     render(<SkillsExplorerTab />);
     await switchToInstalled();
@@ -219,12 +219,12 @@ describe('SkillsExplorerTab', () => {
   });
 
   it('shows skill warnings when present', async () => {
-    const { skillsApi } = await import('../../../services/api/skillsApi');
+    const { workflowsApi } = await import('../../../services/api/workflowsApi');
     const skillWithWarning = {
       ...MOCK_SKILL,
       warnings: ['Missing required field: author'],
     };
-    vi.mocked(skillsApi.listSkills).mockResolvedValue([skillWithWarning]);
+    vi.mocked(workflowsApi.listWorkflows).mockResolvedValue([skillWithWarning]);
 
     render(<SkillsExplorerTab />);
     await switchToInstalled();
@@ -235,12 +235,12 @@ describe('SkillsExplorerTab', () => {
   });
 
   it('shows "Installed" badge for already-installed catalog entries', async () => {
-    const { skillsApi } = await import('../../../services/api/skillsApi');
+    const { workflowsApi } = await import('../../../services/api/workflowsApi');
     const { skillRegistryApi } = await import(
       '../../../services/api/skillRegistryApi'
     );
     const installedSkill = { ...MOCK_SKILL, id: 'registry-skill-1' };
-    vi.mocked(skillsApi.listSkills).mockResolvedValue([installedSkill]);
+    vi.mocked(workflowsApi.listWorkflows).mockResolvedValue([installedSkill]);
     vi.mocked(skillRegistryApi.browse).mockResolvedValue([MOCK_CATALOG_ENTRY]);
 
     render(<SkillsExplorerTab />);
@@ -252,8 +252,8 @@ describe('SkillsExplorerTab', () => {
   });
 
   it('has an install from URL button', async () => {
-    const { skillsApi } = await import('../../../services/api/skillsApi');
-    vi.mocked(skillsApi.listSkills).mockResolvedValue([]);
+    const { workflowsApi } = await import('../../../services/api/workflowsApi');
+    vi.mocked(workflowsApi.listWorkflows).mockResolvedValue([]);
 
     render(<SkillsExplorerTab />);
 

app/src/components/skills/__tests__/UninstallSkillConfirmDialog.test.tsx

@@ -4,7 +4,7 @@
  * Verifies:
  * - Renders skill name + on-disk path + destructive confirm copy.
  * - Cancel button fires onClose, does NOT hit the RPC.
- * - Confirm fires `skillsApi.uninstallSkill(name)` and forwards the result
+ * - Confirm fires `workflowsApi.uninstallWorkflow(name)` and forwards the result
  *   to `onUninstalled`, then closes.
  * - RPC error is surfaced inline and the dialog stays open (no onClose).
  * - While in-flight, both buttons disable and Esc no-ops (handled by
@@ -14,15 +14,15 @@ import { fireEvent, render, screen, waitFor } from '@testing-library/react';
 import { beforeEach, describe, expect, it, vi } from 'vitest';
 
 import UninstallSkillConfirmDialog from '../UninstallSkillConfirmDialog';
-import type { SkillSummary } from '../../../services/api/skillsApi';
+import type { WorkflowSummary } from '../../../services/api/workflowsApi';
 
-vi.mock('../../../services/api/skillsApi', () => ({
-  skillsApi: {
-    uninstallSkill: vi.fn(),
+vi.mock('../../../services/api/workflowsApi', () => ({
+  workflowsApi: {
+    uninstallWorkflow: vi.fn(),
   },
 }));
 
-const fixture: SkillSummary = {
+const fixture: WorkflowSummary = {
   id: 'weather-helper',
   name: 'weather-helper',
   description: 'Weather forecasts',
@@ -43,8 +43,8 @@ const fixture: SkillSummary = {
 
 describe('UninstallSkillConfirmDialog', () => {
   beforeEach(async () => {
-    const { skillsApi } = await import('../../../services/api/skillsApi');
-    vi.mocked(skillsApi.uninstallSkill).mockReset();
+    const { workflowsApi } = await import('../../../services/api/workflowsApi');
+    vi.mocked(workflowsApi.uninstallWorkflow).mockReset();
   });
 
   it('renders skill name, path (stripped of /SKILL.md), and confirm copy', () => {
@@ -68,14 +68,14 @@ describe('UninstallSkillConfirmDialog', () => {
     // by slug — the UI must pass `skill.id` (the slug).
     const onClose = vi.fn();
     const onUninstalled = vi.fn();
-    const { skillsApi } = await import('../../../services/api/skillsApi');
-    vi.mocked(skillsApi.uninstallSkill).mockResolvedValueOnce({
+    const { workflowsApi } = await import('../../../services/api/workflowsApi');
+    vi.mocked(workflowsApi.uninstallWorkflow).mockResolvedValueOnce({
       name: 'weather-helper',
       removedPath: '/Users/me/.openhuman/skills/weather-helper',
       scope: 'user',
     });
 
-    const divergent: SkillSummary = {
+    const divergent: WorkflowSummary = {
       ...fixture,
       id: 'weather-helper',
       name: 'Weather Helper (Pro)',
@@ -90,14 +90,14 @@ describe('UninstallSkillConfirmDialog', () => {
     fireEvent.click(screen.getByTestId('uninstall-skill-confirm'));
 
     await waitFor(() => {
-      expect(vi.mocked(skillsApi.uninstallSkill)).toHaveBeenCalledWith('weather-helper');
+      expect(vi.mocked(workflowsApi.uninstallWorkflow)).toHaveBeenCalledWith('weather-helper');
     });
-    expect(vi.mocked(skillsApi.uninstallSkill)).not.toHaveBeenCalledWith('Weather Helper (Pro)');
+    expect(vi.mocked(workflowsApi.uninstallWorkflow)).not.toHaveBeenCalledWith('Weather Helper (Pro)');
   });
 
   it('Cancel fires onClose without calling the RPC', async () => {
     const onClose = vi.fn();
-    const { skillsApi } = await import('../../../services/api/skillsApi');
+    const { workflowsApi } = await import('../../../services/api/workflowsApi');
     render(
       <UninstallSkillConfirmDialog
         skill={fixture}
@@ -107,14 +107,14 @@ describe('UninstallSkillConfirmDialog', () => {
     );
     fireEvent.click(screen.getByRole('button', { name: /Cancel/ }));
     expect(onClose).toHaveBeenCalledTimes(1);
-    expect(vi.mocked(skillsApi.uninstallSkill)).not.toHaveBeenCalled();
+    expect(vi.mocked(workflowsApi.uninstallWorkflow)).not.toHaveBeenCalled();
   });
 
-  it('Confirm calls skillsApi.uninstallSkill and forwards result to onUninstalled', async () => {
+  it('Confirm calls workflowsApi.uninstallWorkflow and forwards result to onUninstalled', async () => {
     const onClose = vi.fn();
     const onUninstalled = vi.fn();
-    const { skillsApi } = await import('../../../services/api/skillsApi');
-    vi.mocked(skillsApi.uninstallSkill).mockResolvedValueOnce({
+    const { workflowsApi } = await import('../../../services/api/workflowsApi');
+    vi.mocked(workflowsApi.uninstallWorkflow).mockResolvedValueOnce({
       name: 'weather-helper',
       removedPath: '/Users/me/.openhuman/skills/weather-helper',
       scope: 'user',
@@ -130,12 +130,12 @@ describe('UninstallSkillConfirmDialog', () => {
     fireEvent.click(screen.getByTestId('uninstall-skill-confirm'));
 
     await waitFor(() => {
-      expect(vi.mocked(skillsApi.uninstallSkill)).toHaveBeenCalledWith('weather-helper');
+      expect(vi.mocked(workflowsApi.uninstallWorkflow)).toHaveBeenCalledWith('weather-helper');
     });
     // Assert the caller passed the slug (`id`) — not the frontmatter
     // display name. Regression guard for the #781 fix that swapped
     // `skill.name` → `skill.id` in the confirm handler.
-    expect(vi.mocked(skillsApi.uninstallSkill)).toHaveBeenCalledWith(fixture.id);
+    expect(vi.mocked(workflowsApi.uninstallWorkflow)).toHaveBeenCalledWith(fixture.id);
     await waitFor(() => {
       expect(onUninstalled).toHaveBeenCalledWith({
         name: 'weather-helper',
@@ -151,8 +151,8 @@ describe('UninstallSkillConfirmDialog', () => {
   it('surfaces RPC errors inline and keeps the dialog open', async () => {
     const onClose = vi.fn();
     const onUninstalled = vi.fn();
-    const { skillsApi } = await import('../../../services/api/skillsApi');
-    vi.mocked(skillsApi.uninstallSkill).mockRejectedValueOnce(
+    const { workflowsApi } = await import('../../../services/api/workflowsApi');
+    vi.mocked(workflowsApi.uninstallWorkflow).mockRejectedValueOnce(
       new Error("skill 'weather-helper' is not installed")
     );
 
@@ -177,18 +177,18 @@ describe('UninstallSkillConfirmDialog', () => {
   });
 
   it('disables buttons while the RPC is in flight', async () => {
-    const { skillsApi } = await import('../../../services/api/skillsApi');
+    const { workflowsApi } = await import('../../../services/api/workflowsApi');
     type UninstallResolve = (v: {
       name: string;
       removedPath: string;
-      scope: SkillSummary['scope'];
+      scope: WorkflowSummary['scope'];
     }) => void;
     const deferred: { resolve?: UninstallResolve } = {};
-    vi.mocked(skillsApi.uninstallSkill).mockReturnValueOnce(
+    vi.mocked(workflowsApi.uninstallWorkflow).mockReturnValueOnce(
       new Promise<{
         name: string;
         removedPath: string;
-        scope: SkillSummary['scope'];
+        scope: WorkflowSummary['scope'];
       }>(resolve => {
         deferred.resolve = resolve;
       })

app/src/components/skills/__tests__/WorkflowRunnerBody.test.tsx

@@ -9,7 +9,7 @@
  *
  * Covered here:
  *  - Mount with one saved schedule for the picked skill (mocking
- *    skills_list, skills_describe, cron_list, recent_runs).
+ *    workflows_list, workflows_describe, cron_list, recent_runs).
  *  - Toggle flips enabled → false via openhumanCronUpdate(id, { enabled }).
  *  - The list re-loads after toggle (openhumanCronList called again).
  *  - aria-checked reflects the new state once the list refreshes.
@@ -32,9 +32,9 @@ const hoisted = vi.hoisted(() => ({
   cronRun: vi.fn(),
   cronUpdate: vi.fn(),
   cronRuns: vi.fn(),
-  listSkills: vi.fn(),
-  describeSkill: vi.fn(),
-  runSkill: vi.fn(),
+  listWorkflows: vi.fn(),
+  describeWorkflow: vi.fn(),
+  runWorkflow: vi.fn(),
   recentRuns: vi.fn(),
   readRunLog: vi.fn(),
   cancelRun: vi.fn(),
@@ -50,11 +50,11 @@ vi.mock('../../../utils/tauriCommands/cron', () => ({
   openhumanCronRuns: hoisted.cronRuns,
 }));
 
-vi.mock('../../../services/api/skillsApi', () => ({
-  skillsApi: {
-    listSkills: hoisted.listSkills,
-    describeSkill: hoisted.describeSkill,
-    runSkill: hoisted.runSkill,
+vi.mock('../../../services/api/workflowsApi', () => ({
+  workflowsApi: {
+    listWorkflows: hoisted.listWorkflows,
+    describeWorkflow: hoisted.describeWorkflow,
+    runWorkflow: hoisted.runWorkflow,
     recentRuns: hoisted.recentRuns,
     readRunLog: hoisted.readRunLog,
     cancelRun: hoisted.cancelRun,
@@ -158,8 +158,8 @@ describe('WorkflowRunnerBody — saved-schedule toggle', () => {
   beforeEach(() => {
     Object.values(hoisted).forEach((fn) => fn.mockReset());
 
-    hoisted.listSkills.mockResolvedValue(skillsList);
-    hoisted.describeSkill.mockResolvedValue(skillDescription);
+    hoisted.listWorkflows.mockResolvedValue(skillsList);
+    hoisted.describeWorkflow.mockResolvedValue(skillDescription);
     hoisted.recentRuns.mockResolvedValue([]);
     hoisted.cronList.mockResolvedValue({ result: [makeJob({ enabled: true })] });
     hoisted.cronUpdate.mockResolvedValue({ result: makeJob({ enabled: false }) });
@@ -170,8 +170,8 @@ describe('WorkflowRunnerBody — saved-schedule toggle', () => {
     const Body = await importBody();
     renderBody(Body);
 
-    // Wait for skills_list to resolve and populate the dropdown.
-    await waitFor(() => expect(hoisted.listSkills).toHaveBeenCalled());
+    // Wait for workflows_list to resolve and populate the dropdown.
+    await waitFor(() => expect(hoisted.listWorkflows).toHaveBeenCalled());
 
     // Pick the skill so the schedule list mounts.
     const select = screen.getByLabelText('settings.skillsRunner.skill') as HTMLSelectElement;
@@ -192,7 +192,7 @@ describe('WorkflowRunnerBody — saved-schedule toggle', () => {
     const Body = await importBody();
     renderBody(Body);
 
-    await waitFor(() => expect(hoisted.listSkills).toHaveBeenCalled());
+    await waitFor(() => expect(hoisted.listWorkflows).toHaveBeenCalled());
     const select = screen.getByLabelText('settings.skillsRunner.skill') as HTMLSelectElement;
     fireEvent.change(select, { target: { value: SKILL_ID } });
     await waitFor(() => expect(hoisted.cronList).toHaveBeenCalled());
@@ -226,7 +226,7 @@ describe('WorkflowRunnerBody — saved-schedule toggle', () => {
     const Body = await importBody();
     renderBody(Body);
 
-    await waitFor(() => expect(hoisted.listSkills).toHaveBeenCalled());
+    await waitFor(() => expect(hoisted.listWorkflows).toHaveBeenCalled());
     const select = screen.getByLabelText('settings.skillsRunner.skill') as HTMLSelectElement;
     fireEvent.change(select, { target: { value: SKILL_ID } });
     await waitFor(() => expect(hoisted.cronList).toHaveBeenCalled());
@@ -262,8 +262,8 @@ function makeRun(
 describe('WorkflowRunnerBody — per-job history viewer', () => {
   beforeEach(() => {
     Object.values(hoisted).forEach((fn) => fn.mockReset());
-    hoisted.listSkills.mockResolvedValue(skillsList);
-    hoisted.describeSkill.mockResolvedValue(skillDescription);
+    hoisted.listWorkflows.mockResolvedValue(skillsList);
+    hoisted.describeWorkflow.mockResolvedValue(skillDescription);
     hoisted.recentRuns.mockResolvedValue([]);
     hoisted.cronList.mockResolvedValue({ result: [makeJob({ enabled: true })] });
     hoisted.cronRuns.mockResolvedValue({ result: { runs: [makeRun(1), makeRun(2)] } });
@@ -272,7 +272,7 @@ describe('WorkflowRunnerBody — per-job history viewer', () => {
   it('loads cron_runs and renders history rows on first toggle', async () => {
     const Body = await importBody();
     renderBody(Body);
-    await waitFor(() => expect(hoisted.listSkills).toHaveBeenCalled());
+    await waitFor(() => expect(hoisted.listWorkflows).toHaveBeenCalled());
     const select = screen.getByLabelText('settings.skillsRunner.skill') as HTMLSelectElement;
     fireEvent.change(select, { target: { value: SKILL_ID } });
     await waitFor(() => expect(hoisted.cronList).toHaveBeenCalled());
@@ -288,7 +288,7 @@ describe('WorkflowRunnerBody — per-job history viewer', () => {
   it("expands a run row to show its captured output, hides on collapse", async () => {
     const Body = await importBody();
     renderBody(Body);
-    await waitFor(() => expect(hoisted.listSkills).toHaveBeenCalled());
+    await waitFor(() => expect(hoisted.listWorkflows).toHaveBeenCalled());
     const select = screen.getByLabelText('settings.skillsRunner.skill') as HTMLSelectElement;
     fireEvent.change(select, { target: { value: SKILL_ID } });
     await waitFor(() => expect(hoisted.cronList).toHaveBeenCalled());
@@ -329,7 +329,7 @@ describe('WorkflowRunnerBody — per-job history viewer', () => {
 
     const Body = await importBody();
     renderBody(Body);
-    await waitFor(() => expect(hoisted.listSkills).toHaveBeenCalled());
+    await waitFor(() => expect(hoisted.listWorkflows).toHaveBeenCalled());
     const select = screen.getByLabelText('settings.skillsRunner.skill') as HTMLSelectElement;
     fireEvent.change(select, { target: { value: SKILL_ID } });
     await waitFor(() => expect(hoisted.cronList).toHaveBeenCalled());
@@ -374,7 +374,7 @@ describe('WorkflowRunnerBody — per-job history viewer', () => {
     });
     const Body = await importBody();
     renderBody(Body);
-    await waitFor(() => expect(hoisted.listSkills).toHaveBeenCalled());
+    await waitFor(() => expect(hoisted.listWorkflows).toHaveBeenCalled());
     const select = screen.getByLabelText('settings.skillsRunner.skill') as HTMLSelectElement;
     fireEvent.change(select, { target: { value: SKILL_ID } });
     await waitFor(() => expect(hoisted.cronList).toHaveBeenCalled());
@@ -387,7 +387,7 @@ describe('WorkflowRunnerBody — per-job history viewer', () => {
     hoisted.cronRuns.mockResolvedValue({ result: { runs: [] } });
     const Body = await importBody();
     renderBody(Body);
-    await waitFor(() => expect(hoisted.listSkills).toHaveBeenCalled());
+    await waitFor(() => expect(hoisted.listWorkflows).toHaveBeenCalled());
     const select = screen.getByLabelText('settings.skillsRunner.skill') as HTMLSelectElement;
     fireEvent.change(select, { target: { value: SKILL_ID } });
     await waitFor(() => expect(hoisted.cronList).toHaveBeenCalled());
@@ -403,8 +403,8 @@ describe('WorkflowRunnerBody — per-job history viewer', () => {
 describe('WorkflowRunnerBody — schedule frequency + save', () => {
   beforeEach(() => {
     Object.values(hoisted).forEach((fn) => fn.mockReset());
-    hoisted.listSkills.mockResolvedValue(skillsList);
-    hoisted.describeSkill.mockResolvedValue(skillDescription);
+    hoisted.listWorkflows.mockResolvedValue(skillsList);
+    hoisted.describeWorkflow.mockResolvedValue(skillDescription);
     hoisted.recentRuns.mockResolvedValue([]);
     hoisted.cronList.mockResolvedValue({ result: [] });
     hoisted.cronAdd.mockResolvedValue({ result: makeJob() });
@@ -414,7 +414,7 @@ describe('WorkflowRunnerBody — schedule frequency + save', () => {
   it('changes schedule frequency and calls openhumanCronAdd on save', async () => {
     const Body = await importBody();
     renderBody(Body);
-    await waitFor(() => expect(hoisted.listSkills).toHaveBeenCalled());
+    await waitFor(() => expect(hoisted.listWorkflows).toHaveBeenCalled());
     const select = screen.getByLabelText('settings.skillsRunner.skill') as HTMLSelectElement;
     fireEvent.change(select, { target: { value: SKILL_ID } });
     await waitFor(() => expect(hoisted.cronList).toHaveBeenCalled());
@@ -444,8 +444,8 @@ describe('WorkflowRunnerBody — SmartIssuePicker conditional mount', () => {
   });
 
   it('renders SmartIssuePicker when the picked skill is dev-workflow', async () => {
-    hoisted.listSkills.mockResolvedValue([{ id: 'dev-workflow', name: 'Dev Workflow' }]);
-    hoisted.describeSkill.mockResolvedValue({
+    hoisted.listWorkflows.mockResolvedValue([{ id: 'dev-workflow', name: 'Dev Workflow' }]);
+    hoisted.describeWorkflow.mockResolvedValue({
       id: 'dev-workflow',
       name: 'Dev Workflow',
       when_to_use: 'Autonomous developer.',
@@ -459,7 +459,7 @@ describe('WorkflowRunnerBody — SmartIssuePicker conditional mount', () => {
 
     const Body = await importBody();
     renderBody(Body);
-    await waitFor(() => expect(hoisted.listSkills).toHaveBeenCalled());
+    await waitFor(() => expect(hoisted.listWorkflows).toHaveBeenCalled());
     const select = screen.getByLabelText('settings.skillsRunner.skill') as HTMLSelectElement;
     fireEvent.change(select, { target: { value: 'dev-workflow' } });
 
@@ -470,10 +470,10 @@ describe('WorkflowRunnerBody — SmartIssuePicker conditional mount', () => {
   });
 
   it('does NOT render SmartIssuePicker for generic skills', async () => {
-    hoisted.listSkills.mockResolvedValue([
+    hoisted.listWorkflows.mockResolvedValue([
       { id: 'github-issue-crusher', name: 'GitHub Issue Crusher' },
     ]);
-    hoisted.describeSkill.mockResolvedValue({
+    hoisted.describeWorkflow.mockResolvedValue({
       id: 'github-issue-crusher',
       name: 'GitHub Issue Crusher',
       when_to_use: 'Crush issues.',
@@ -485,11 +485,11 @@ describe('WorkflowRunnerBody — SmartIssuePicker conditional mount', () => {
 
     const Body = await importBody();
     renderBody(Body);
-    await waitFor(() => expect(hoisted.listSkills).toHaveBeenCalled());
+    await waitFor(() => expect(hoisted.listWorkflows).toHaveBeenCalled());
     const select = screen.getByLabelText('settings.skillsRunner.skill') as HTMLSelectElement;
     fireEvent.change(select, { target: { value: 'github-issue-crusher' } });
 
-    await waitFor(() => expect(hoisted.describeSkill).toHaveBeenCalled());
+    await waitFor(() => expect(hoisted.describeWorkflow).toHaveBeenCalled());
     expect(screen.queryByTestId('smart-issue-picker-stub')).not.toBeInTheDocument();
     // The generic schema-driven repo field IS rendered via the
     // existing RepoPicker stub.
@@ -502,11 +502,11 @@ describe('WorkflowRunnerBody — SmartIssuePicker conditional mount', () => {
 describe('WorkflowRunnerBody — URL ?workflow= preselect', () => {
   beforeEach(() => {
     Object.values(hoisted).forEach((fn) => fn.mockReset());
-    hoisted.listSkills.mockResolvedValue([
+    hoisted.listWorkflows.mockResolvedValue([
       { id: 'dev-workflow', name: 'Dev Workflow' },
       { id: 'github-issue-crusher', name: 'GitHub Issue Crusher' },
     ]);
-    hoisted.describeSkill.mockResolvedValue({
+    hoisted.describeWorkflow.mockResolvedValue({
       id: 'dev-workflow',
       name: 'Dev Workflow',
       when_to_use: 'Autonomous developer.',
@@ -521,17 +521,17 @@ describe('WorkflowRunnerBody — URL ?workflow= preselect', () => {
     const Body = await importBody();
     renderBody(Body, '/workflows/run?workflow=dev-workflow');
 
-    await waitFor(() => expect(hoisted.listSkills).toHaveBeenCalled());
+    await waitFor(() => expect(hoisted.listWorkflows).toHaveBeenCalled());
 
     // The picker should already be pointing at dev-workflow without any
     // user interaction. We assert this two ways: (a) the <select>'s
-    // value matches, and (b) describeSkill was fetched for it.
+    // value matches, and (b) describeWorkflow was fetched for it.
     const select = (await screen.findByLabelText(
       'settings.skillsRunner.skill'
     )) as HTMLSelectElement;
     expect(select.value).toBe('dev-workflow');
     await waitFor(() =>
-      expect(hoisted.describeSkill).toHaveBeenCalledWith('dev-workflow')
+      expect(hoisted.describeWorkflow).toHaveBeenCalledWith('dev-workflow')
     );
   });
 
@@ -539,27 +539,27 @@ describe('WorkflowRunnerBody — URL ?workflow= preselect', () => {
     const Body = await importBody();
     renderBody(Body, '/workflows/run');
 
-    await waitFor(() => expect(hoisted.listSkills).toHaveBeenCalled());
+    await waitFor(() => expect(hoisted.listWorkflows).toHaveBeenCalled());
     const select = (await screen.findByLabelText(
       'settings.skillsRunner.skill'
     )) as HTMLSelectElement;
     expect(select.value).toBe('');
-    expect(hoisted.describeSkill).not.toHaveBeenCalled();
+    expect(hoisted.describeWorkflow).not.toHaveBeenCalled();
   });
 
-  it('ignores ?workflow= when the value is not in the skills_list (picker stays empty, describeSkill called once with empty=never)', async () => {
+  it('ignores ?workflow= when the value is not in the workflows_list (picker stays empty, describeWorkflow called once with empty=never)', async () => {
     // ?workflow=unknown-skill is treated as best-effort: we set the state
     // but the picker shows "Select a skill" since the option isn't in
     // the list. The describe call IS attempted (we don't pre-filter
     // against the catalog) — but the cancellation effect tears it
     // down if the value never resolves to a real skill.
-    hoisted.describeSkill.mockRejectedValue(new Error('unknown skill'));
+    hoisted.describeWorkflow.mockRejectedValue(new Error('unknown skill'));
     const Body = await importBody();
     renderBody(Body, '/workflows/run?workflow=does-not-exist');
 
-    await waitFor(() => expect(hoisted.listSkills).toHaveBeenCalled());
+    await waitFor(() => expect(hoisted.listWorkflows).toHaveBeenCalled());
     await waitFor(() =>
-      expect(hoisted.describeSkill).toHaveBeenCalledWith('does-not-exist')
+      expect(hoisted.describeWorkflow).toHaveBeenCalledWith('does-not-exist')
     );
     // The dropdown value won't render as an option (not in the list),
     // so its current value normalises to '' visually — but the state
@@ -576,7 +576,7 @@ describe('WorkflowRunnerBody — URL ?workflow= preselect', () => {
     const Body = await importBody();
     renderBody(Body, '/workflows/run?workflow=dev-workflow&lock=1');
 
-    await waitFor(() => expect(hoisted.listSkills).toHaveBeenCalled());
+    await waitFor(() => expect(hoisted.listWorkflows).toHaveBeenCalled());
 
     // The picker is replaced by the workflow-name heading...
     const heading = await screen.findByTestId('skills-runner-skill-locked');
@@ -601,8 +601,8 @@ describe('WorkflowRunnerBody — URL ?workflow= preselect', () => {
 describe('WorkflowRunnerBody — Run Now flow', () => {
   beforeEach(() => {
     Object.values(hoisted).forEach((fn) => fn.mockReset());
-    hoisted.listSkills.mockResolvedValue([{ id: 'pr-review-shepherd', name: 'PR Review Shepherd' }]);
-    hoisted.describeSkill.mockResolvedValue({
+    hoisted.listWorkflows.mockResolvedValue([{ id: 'pr-review-shepherd', name: 'PR Review Shepherd' }]);
+    hoisted.describeWorkflow.mockResolvedValue({
       id: 'pr-review-shepherd',
       name: 'PR Review Shepherd',
       when_to_use: 'Shepherd PRs.',
@@ -615,7 +615,7 @@ describe('WorkflowRunnerBody — Run Now flow', () => {
     hoisted.recentRuns.mockResolvedValue([]);
     hoisted.cronList.mockResolvedValue({ result: [] });
     hoisted.cronRuns.mockResolvedValue({ result: { runs: [] } });
-    hoisted.runSkill.mockResolvedValue({
+    hoisted.runWorkflow.mockResolvedValue({
       run_id: 'run-abc',
       skill_id: 'pr-review-shepherd',
       log: '/tmp/run-abc.log',
@@ -625,26 +625,26 @@ describe('WorkflowRunnerBody — Run Now flow', () => {
   it('Run Now button is disabled while required fields are empty', async () => {
     const Body = await importBody();
     renderBody(Body);
-    await waitFor(() => expect(hoisted.listSkills).toHaveBeenCalled());
+    await waitFor(() => expect(hoisted.listWorkflows).toHaveBeenCalled());
 
     const select = screen.getByLabelText('settings.skillsRunner.skill') as HTMLSelectElement;
     fireEvent.change(select, { target: { value: 'pr-review-shepherd' } });
-    await waitFor(() => expect(hoisted.describeSkill).toHaveBeenCalledWith('pr-review-shepherd'));
+    await waitFor(() => expect(hoisted.describeWorkflow).toHaveBeenCalledWith('pr-review-shepherd'));
 
     // Run Now button should be disabled when required field is empty
     const runBtn = await screen.findByText('settings.skillsRunner.runNow');
     expect(runBtn.closest('button')).toBeDisabled();
-    expect(hoisted.runSkill).not.toHaveBeenCalled();
+    expect(hoisted.runWorkflow).not.toHaveBeenCalled();
   });
 
-  it('calls runSkill with built payload when required fields are filled', async () => {
+  it('calls runWorkflow with built payload when required fields are filled', async () => {
     const Body = await importBody();
     renderBody(Body);
-    await waitFor(() => expect(hoisted.listSkills).toHaveBeenCalled());
+    await waitFor(() => expect(hoisted.listWorkflows).toHaveBeenCalled());
 
     const select = screen.getByLabelText('settings.skillsRunner.skill') as HTMLSelectElement;
     fireEvent.change(select, { target: { value: 'pr-review-shepherd' } });
-    await waitFor(() => expect(hoisted.describeSkill).toHaveBeenCalled());
+    await waitFor(() => expect(hoisted.describeWorkflow).toHaveBeenCalled());
 
     // Fill the repo input (rendered as a RepoPicker stub <input>)
     const repoInput = await screen.findByTestId('repo-picker-stub');
@@ -656,7 +656,7 @@ describe('WorkflowRunnerBody — Run Now flow', () => {
     fireEvent.click(runBtn.closest('button')!);
 
     await waitFor(() =>
-      expect(hoisted.runSkill).toHaveBeenCalledWith(
+      expect(hoisted.runWorkflow).toHaveBeenCalledWith(
         'pr-review-shepherd',
         expect.objectContaining({ repo: 'owner/myrepo' })
       )
@@ -669,11 +669,11 @@ describe('WorkflowRunnerBody — Run Now flow', () => {
     // click Run again and spawn a second run.
     const Body = await importBody();
     renderBody(Body);
-    await waitFor(() => expect(hoisted.listSkills).toHaveBeenCalled());
+    await waitFor(() => expect(hoisted.listWorkflows).toHaveBeenCalled());
 
     const select = screen.getByLabelText('settings.skillsRunner.skill') as HTMLSelectElement;
     fireEvent.change(select, { target: { value: 'pr-review-shepherd' } });
-    await waitFor(() => expect(hoisted.describeSkill).toHaveBeenCalled());
+    await waitFor(() => expect(hoisted.describeWorkflow).toHaveBeenCalled());
 
     const repoInput = await screen.findByTestId('repo-picker-stub');
     fireEvent.change(repoInput, { target: { value: 'owner/myrepo' } });
@@ -684,22 +684,22 @@ describe('WorkflowRunnerBody — Run Now flow', () => {
     const callsBefore = hoisted.recentRuns.mock.calls.length;
     fireEvent.click(runBtn.closest('button')!);
 
-    await waitFor(() => expect(hoisted.runSkill).toHaveBeenCalledTimes(1));
+    await waitFor(() => expect(hoisted.runWorkflow).toHaveBeenCalledTimes(1));
     // The post-run refresh burst re-scans recentRuns on its own.
     await waitFor(() =>
       expect(hoisted.recentRuns.mock.calls.length).toBeGreaterThan(callsBefore)
     );
   });
 
-  it('surfaces error when runSkill rejects', async () => {
-    hoisted.runSkill.mockRejectedValue(new Error('backend error'));
+  it('surfaces error when runWorkflow rejects', async () => {
+    hoisted.runWorkflow.mockRejectedValue(new Error('backend error'));
     const Body = await importBody();
     renderBody(Body);
-    await waitFor(() => expect(hoisted.listSkills).toHaveBeenCalled());
+    await waitFor(() => expect(hoisted.listWorkflows).toHaveBeenCalled());
 
     const select = screen.getByLabelText('settings.skillsRunner.skill') as HTMLSelectElement;
     fireEvent.change(select, { target: { value: 'pr-review-shepherd' } });
-    await waitFor(() => expect(hoisted.describeSkill).toHaveBeenCalled());
+    await waitFor(() => expect(hoisted.describeWorkflow).toHaveBeenCalled());
 
     const repoInput = await screen.findByTestId('repo-picker-stub');
     fireEvent.change(repoInput, { target: { value: 'owner/myrepo' } });
@@ -762,14 +762,14 @@ describe('parseScheduledInputs', () => {
 describe('WorkflowRunnerBody — Stop / Edit / scheduled run-now', () => {
   beforeEach(() => {
     Object.values(hoisted).forEach((fn) => fn.mockReset());
-    hoisted.listSkills.mockResolvedValue([
+    hoisted.listWorkflows.mockResolvedValue([
       { id: SKILL_ID, name: 'GitHub Issue Crusher', scope: 'user', legacy: false },
     ]);
-    hoisted.describeSkill.mockResolvedValue(skillDescription);
+    hoisted.describeWorkflow.mockResolvedValue(skillDescription);
     hoisted.recentRuns.mockResolvedValue([]);
     hoisted.cronList.mockResolvedValue({ result: [] });
     hoisted.cronRuns.mockResolvedValue({ result: { runs: [] } });
-    hoisted.runSkill.mockResolvedValue({ run_id: 'r-new', workflow_id: SKILL_ID, log: '/tmp/l' });
+    hoisted.runWorkflow.mockResolvedValue({ run_id: 'r-new', workflow_id: SKILL_ID, log: '/tmp/l' });
     hoisted.cancelRun.mockResolvedValue(true);
   });
 
@@ -820,7 +820,7 @@ describe('WorkflowRunnerBody — Stop / Edit / scheduled run-now', () => {
     // The card's "Run" runs the workflow directly with those inputs.
     fireEvent.click(screen.getByText('settings.skillsRunner.schedule.runNow'));
     await waitFor(() =>
-      expect(hoisted.runSkill).toHaveBeenCalledWith(SKILL_ID, { channel: 'team-product' })
+      expect(hoisted.runWorkflow).toHaveBeenCalledWith(SKILL_ID, { channel: 'team-product' })
     );
   });
 });

app/src/components/skills/preflightGate.test.ts

@@ -1,25 +1,25 @@
 import { describe, expect, it } from 'vitest';
 
-import { isGithubGateFailure, parseSkillRunError } from './preflightGate';
+import { isGithubGateFailure, parseWorkflowRunError } from './preflightGate';
 
-describe('parseSkillRunError', () => {
+describe('parseWorkflowRunError', () => {
   it('returns the raw body unchanged when no preflight prefix is present', () => {
-    const out = parseSkillRunError('Run failed because foo');
+    const out = parseWorkflowRunError('Run failed because foo');
     expect(out.gate).toBeNull();
     expect(out.tag).toBeNull();
     expect(out.body).toBe('Run failed because foo');
   });
 
   it('handles null / undefined / empty without throwing', () => {
-    expect(parseSkillRunError(null).body).toBe('');
-    expect(parseSkillRunError(undefined).body).toBe('');
-    expect(parseSkillRunError('').body).toBe('');
+    expect(parseWorkflowRunError(null).body).toBe('');
+    expect(parseWorkflowRunError(undefined).body).toBe('');
+    expect(parseWorkflowRunError('').body).toBe('');
   });
 
   it('parses a github identity_mismatch failure into gate + tag + body', () => {
     const raw =
       '[preflight:github:identity_mismatch] GitHub preflight failed: identity mismatch — Composio is `octo-alice` but git is `Alice`.';
-    const out = parseSkillRunError(raw);
+    const out = parseWorkflowRunError(raw);
     expect(out.gate).toBe('github');
     expect(out.tag).toBe('identity_mismatch');
     expect(out.body).toContain('GitHub preflight failed');
@@ -37,7 +37,7 @@ describe('parseSkillRunError', () => {
     ];
     for (const tag of tags) {
       const raw = `[preflight:github:${tag}] body for ${tag}`;
-      const out = parseSkillRunError(raw);
+      const out = parseWorkflowRunError(raw);
       expect(out.gate).toBe('github');
       expect(out.tag).toBe(tag);
       expect(out.body).toBe(`body for ${tag}`);
@@ -46,15 +46,15 @@ describe('parseSkillRunError', () => {
 
   it('is idempotent — re-parsing a stripped body is a no-op', () => {
     const raw = '[preflight:github:git_user_name_missing] please set user.name';
-    const once = parseSkillRunError(raw);
-    const twice = parseSkillRunError(once.body);
+    const once = parseWorkflowRunError(raw);
+    const twice = parseWorkflowRunError(once.body);
     expect(twice.gate).toBeNull();
     expect(twice.tag).toBeNull();
     expect(twice.body).toBe(once.body);
   });
 
   it('tolerates lowercase / mixed case in the prefix gate name', () => {
-    const out = parseSkillRunError('[preflight:GITHUB:Identity_Mismatch] body');
+    const out = parseWorkflowRunError('[preflight:GITHUB:Identity_Mismatch] body');
     expect(out.gate).toBe('github');
     expect(out.tag).toBe('identity_mismatch');
     expect(out.body).toBe('body');
@@ -64,7 +64,7 @@ describe('parseSkillRunError', () => {
     // Anchored at start of string only — a dump that contains the
     // prefix mid-text shouldn't be misinterpreted.
     const raw = 'orchestrator log: [preflight:github:tag] not the head of the string';
-    const out = parseSkillRunError(raw);
+    const out = parseWorkflowRunError(raw);
     expect(out.gate).toBeNull();
     expect(out.body).toBe(raw);
   });
@@ -72,17 +72,17 @@ describe('parseSkillRunError', () => {
 
 describe('isGithubGateFailure', () => {
   it('returns true for a github-gate parsed error', () => {
-    const err = parseSkillRunError('[preflight:github:identity_mismatch] x');
+    const err = parseWorkflowRunError('[preflight:github:identity_mismatch] x');
     expect(isGithubGateFailure(err)).toBe(true);
   });
 
   it('returns false for a free-form error', () => {
-    const err = parseSkillRunError('Something else failed');
+    const err = parseWorkflowRunError('Something else failed');
     expect(isGithubGateFailure(err)).toBe(false);
   });
 
   it('returns false for a future non-github gate (forward-compat)', () => {
-    const err = parseSkillRunError('[preflight:slack:scope_missing] body');
+    const err = parseWorkflowRunError('[preflight:slack:scope_missing] body');
     expect(isGithubGateFailure(err)).toBe(false);
     expect(err.gate).toBe('slack');
   });

app/src/components/skills/preflightGate.ts

@@ -23,7 +23,7 @@
 const PREFLIGHT_PREFIX_RE = /^\[preflight:([a-z0-9_-]+):([a-z0-9_-]+)\]\s+/i;
 
 /** Parsed shape of a backend RPC error returned by `openhuman.workflows_run`. */
-export interface SkillRunError {
+export interface WorkflowRunError {
   /** `'github'` when this is a github-gate failure; `null` for any other error. */
   gate: string | null;
   /**
@@ -37,7 +37,7 @@ export interface SkillRunError {
 
 /**
  * Parse the message string returned by the `openhuman.workflows_run` RPC
- * error path (or thrown by `skillsApi.runSkill`). Anything that matches
+ * error path (or thrown by `workflowsApi.runWorkflow`). Anything that matches
  * the `[preflight:<gate>:<tag>]` prefix becomes a structured gate
  * failure; anything else falls through with `gate: null` so the caller
  * can render the raw text.
@@ -45,7 +45,7 @@ export interface SkillRunError {
  * Idempotent: calling this twice on the same message is fine (the
  * second call sees no prefix and returns the same body unchanged).
  */
-export function parseSkillRunError(message: string | undefined | null): SkillRunError {
+export function parseWorkflowRunError(message: string | undefined | null): WorkflowRunError {
   const raw = (message ?? '').toString();
   const m = PREFLIGHT_PREFIX_RE.exec(raw);
   if (!m) {
@@ -63,6 +63,6 @@ export function parseSkillRunError(message: string | undefined | null): SkillRun
  * the rendering layer that wants a single boolean rather than
  * matching on the `gate` string.
  */
-export function isGithubGateFailure(err: SkillRunError): boolean {
+export function isGithubGateFailure(err: WorkflowRunError): boolean {
   return err.gate === 'github';
 }

app/src/pages/WorkflowNew.test.tsx

@@ -5,7 +5,7 @@
  *  - renders the form (delegates to CreateWorkflowForm) and the header
  *    Cancel/Submit buttons.
  *  - cancel button navigates back to /skills.
- *  - on a successful submit (createSkill resolves), the page
+ *  - on a successful submit (createWorkflow resolves), the page
  *    navigates to /skills.
  *  - submit button reflects the form's validity (disabled until both
  *    required fields are filled).
@@ -19,9 +19,11 @@ import WorkflowNew from './WorkflowNew';
 const stableT = (key: string) => key;
 vi.mock('../lib/i18n/I18nContext', () => ({ useT: () => ({ t: stableT }) }));
 
-const hoisted = vi.hoisted(() => ({ createSkill: vi.fn() }));
+const hoisted = vi.hoisted(() => ({ createWorkflow: vi.fn() }));
 
-vi.mock('../services/api/skillsApi', () => ({ skillsApi: { createSkill: hoisted.createSkill } }));
+vi.mock('../services/api/workflowsApi', () => ({
+  workflowsApi: { createWorkflow: hoisted.createWorkflow },
+}));
 
 const renderPage = () =>
   render(
@@ -35,7 +37,7 @@ const renderPage = () =>
 
 describe('WorkflowNew', () => {
   beforeEach(() => {
-    hoisted.createSkill.mockReset();
+    hoisted.createWorkflow.mockReset();
   });
 
   it('renders the form and the header CTAs', () => {
@@ -70,7 +72,7 @@ describe('WorkflowNew', () => {
   });
 
   it('navigates to /skills after a successful submit', async () => {
-    hoisted.createSkill.mockResolvedValue({
+    hoisted.createWorkflow.mockResolvedValue({
       id: 'new-skill',
       name: 'New Skill',
       scope: 'user',
@@ -86,7 +88,7 @@ describe('WorkflowNew', () => {
     });
 
     fireEvent.click(screen.getByTestId('skill-new-submit'));
-    await waitFor(() => expect(hoisted.createSkill).toHaveBeenCalled());
+    await waitFor(() => expect(hoisted.createWorkflow).toHaveBeenCalled());
     await screen.findByTestId('dashboard-landed');
   });
 });

app/src/pages/WorkflowNew.tsx

@@ -21,7 +21,7 @@ import { useNavigate } from 'react-router-dom';
 
 import CreateWorkflowForm from '../components/skills/CreateWorkflowForm';
 import { useT } from '../lib/i18n/I18nContext';
-import { type SkillSummary } from '../services/api/skillsApi';
+import { type WorkflowSummary } from '../services/api/workflowsApi';
 
 const PAGE_FORM_ID = 'create-skill-page-form';
 
@@ -38,7 +38,7 @@ export default function WorkflowNew() {
   }, []);
 
   const handleCreated = useCallback(
-    (_skill: SkillSummary) => {
+    (_skill: WorkflowSummary) => {
       // The dashboard re-fetches the cron list on mount, so any
       // schedule the user adds for this new skill will appear there
       // automatically — no need to plumb the new id through state.

app/src/pages/__tests__/Skills.channels-grid.test.tsx

@@ -32,13 +32,13 @@ vi.mock('../../hooks/useChannelDefinitions', () => ({
   }),
 }));
 
-vi.mock('../../services/api/skillsApi', async () => {
-  const actual = await vi.importActual<typeof import('../../services/api/skillsApi')>(
-    '../../services/api/skillsApi'
+vi.mock('../../services/api/workflowsApi', async () => {
+  const actual = await vi.importActual<typeof import('../../services/api/workflowsApi')>(
+    '../../services/api/workflowsApi'
   );
   return {
     ...actual,
-    skillsApi: { ...actual.skillsApi, listSkills: vi.fn().mockResolvedValue([]) },
+    workflowsApi: { ...actual.workflowsApi, listWorkflows: vi.fn().mockResolvedValue([]) },
   };
 });
 

app/src/pages/__tests__/Skills.mcp-coming-soon.test.tsx

@@ -9,13 +9,13 @@ vi.mock('../../hooks/useChannelDefinitions', () => ({
   useChannelDefinitions: () => ({ definitions: [], loading: false, error: null }),
 }));
 
-vi.mock('../../services/api/skillsApi', async () => {
-  const actual = await vi.importActual<typeof import('../../services/api/skillsApi')>(
-    '../../services/api/skillsApi'
+vi.mock('../../services/api/workflowsApi', async () => {
+  const actual = await vi.importActual<typeof import('../../services/api/workflowsApi')>(
+    '../../services/api/workflowsApi'
   );
   return {
     ...actual,
-    skillsApi: { ...actual.skillsApi, listSkills: vi.fn().mockResolvedValue([]) },
+    workflowsApi: { ...actual.workflowsApi, listWorkflows: vi.fn().mockResolvedValue([]) },
   };
 });
 

app/src/pages/__tests__/Skills.meetings-tab.test.tsx

@@ -13,13 +13,13 @@ vi.mock('../../hooks/useChannelDefinitions', () => ({
   useChannelDefinitions: () => ({ definitions: [], loading: false, error: null }),
 }));
 
-vi.mock('../../services/api/skillsApi', async () => {
-  const actual = await vi.importActual<typeof import('../../services/api/skillsApi')>(
-    '../../services/api/skillsApi'
+vi.mock('../../services/api/workflowsApi', async () => {
+  const actual = await vi.importActual<typeof import('../../services/api/workflowsApi')>(
+    '../../services/api/workflowsApi'
   );
   return {
     ...actual,
-    skillsApi: { ...actual.skillsApi, listSkills: vi.fn().mockResolvedValue([]) },
+    workflowsApi: { ...actual.workflowsApi, listWorkflows: vi.fn().mockResolvedValue([]) },
   };
 });
 

app/src/services/api/__tests__/workflowsApi.test.ts

@@ -1,19 +1,243 @@
 import { beforeEach, describe, expect, it, vi } from 'vitest';
 
-import { skillsApi } from '../skillsApi';
+import { workflowsApi } from '../workflowsApi';
 
 vi.mock('../../coreRpcClient', () => ({ callCoreRpc: vi.fn() }));
 
-describe('skillsApi.listSkills', () => {
+describe('workflowsApi.createWorkflow', () => {
   beforeEach(async () => {
     const { callCoreRpc } = await import('../../coreRpcClient');
     vi.mocked(callCoreRpc).mockReset();
   });
 
-  it('normalizes snake_case and legacy discovered skill fields', async () => {
+  it('forwards inputs to workflows_create and rekeys allowedTools', async () => {
     const { callCoreRpc } = await import('../../coreRpcClient');
     vi.mocked(callCoreRpc).mockResolvedValueOnce({
-      skills: [
+      workflow: {
+        id: 'my-skill',
+        name: 'my-skill',
+        description: 'does stuff',
+        version: '',
+        author: null,
+        tags: ['alpha'],
+        tools: ['mcp/fs'],
+        prompts: [],
+        location: '/home/u/.openhuman/skills/my-skill/SKILL.md',
+        resources: [],
+        scope: 'user',
+        legacy: false,
+        warnings: [],
+      },
+    });
+
+    const result = await workflowsApi.createWorkflow({
+      name: 'My Skill',
+      description: 'does stuff',
+      scope: 'user',
+      tags: ['alpha'],
+      allowedTools: ['mcp/fs'],
+    });
+
+    expect(callCoreRpc).toHaveBeenCalledWith({
+      method: 'openhuman.workflows_create',
+      params: {
+        name: 'My Skill',
+        description: 'does stuff',
+        scope: 'user',
+        tags: ['alpha'],
+        'allowed-tools': ['mcp/fs'],
+      },
+    });
+    expect(result.id).toBe('my-skill');
+    expect(result.scope).toBe('user');
+  });
+
+  it('omits optional fields when not provided', async () => {
+    const { callCoreRpc } = await import('../../coreRpcClient');
+    vi.mocked(callCoreRpc).mockResolvedValueOnce({
+      workflow: {
+        id: 'minimal',
+        name: 'minimal',
+        description: 'd',
+        version: '',
+        author: null,
+        tags: [],
+        tools: [],
+        prompts: [],
+        location: null,
+        resources: [],
+        scope: 'user',
+        legacy: false,
+        warnings: [],
+      },
+    });
+
+    await workflowsApi.createWorkflow({ name: 'minimal', description: 'd' });
+
+    const call = vi.mocked(callCoreRpc).mock.calls[0][0];
+    expect(call.params).toEqual({ name: 'minimal', description: 'd' });
+  });
+
+  it('unwraps an envelope response', async () => {
+    const { callCoreRpc } = await import('../../coreRpcClient');
+    vi.mocked(callCoreRpc).mockResolvedValueOnce({
+      data: {
+        workflow: {
+          id: 'env',
+          name: 'env',
+          description: 'e',
+          version: '',
+          author: null,
+          tags: [],
+          tools: [],
+          prompts: [],
+          location: null,
+          resources: [],
+          scope: 'project',
+          legacy: false,
+          warnings: [],
+        },
+      },
+    });
+    const result = await workflowsApi.createWorkflow({ name: 'env', description: 'e' });
+    expect(result.id).toBe('env');
+    expect(result.scope).toBe('project');
+  });
+});
+
+describe('workflowsApi.installWorkflowFromUrl', () => {
+  beforeEach(async () => {
+    const { callCoreRpc } = await import('../../coreRpcClient');
+    vi.mocked(callCoreRpc).mockReset();
+  });
+
+  it('forwards url and rekeys timeoutSecs to timeout_secs', async () => {
+    const { callCoreRpc } = await import('../../coreRpcClient');
+    vi.mocked(callCoreRpc).mockResolvedValueOnce({
+      url: 'https://example.com/my-skill.tgz',
+      stdout: 'added my-skill',
+      stderr: '',
+      new_workflows: ['my-skill'],
+    });
+
+    const result = await workflowsApi.installWorkflowFromUrl({
+      url: 'https://example.com/my-skill.tgz',
+      timeoutSecs: 120,
+    });
+
+    expect(callCoreRpc).toHaveBeenCalledWith({
+      method: 'openhuman.workflows_install_from_url',
+      params: { url: 'https://example.com/my-skill.tgz', timeout_secs: 120 },
+    });
+    expect(result.newWorkflows).toEqual(['my-skill']);
+    expect(result.stdout).toBe('added my-skill');
+  });
+
+  it('omits timeout_secs when not provided and normalizes missing new_workflows', async () => {
+    const { callCoreRpc } = await import('../../coreRpcClient');
+    vi.mocked(callCoreRpc).mockResolvedValueOnce({
+      url: 'https://example.com/x',
+      stdout: '',
+      stderr: '',
+      new_workflows: undefined,
+    });
+
+    const result = await workflowsApi.installWorkflowFromUrl({ url: 'https://example.com/x' });
+
+    const call = vi.mocked(callCoreRpc).mock.calls[0][0];
+    expect(call.params).toEqual({ url: 'https://example.com/x' });
+    expect(result.newWorkflows).toEqual([]);
+  });
+
+  it('unwraps an envelope response', async () => {
+    const { callCoreRpc } = await import('../../coreRpcClient');
+    vi.mocked(callCoreRpc).mockResolvedValueOnce({
+      data: {
+        url: 'https://example.com/y',
+        stdout: 'ok',
+        stderr: 'warn',
+        new_workflows: ['y-skill'],
+      },
+    });
+    const result = await workflowsApi.installWorkflowFromUrl({ url: 'https://example.com/y' });
+    expect(result.newWorkflows).toEqual(['y-skill']);
+    expect(result.stderr).toBe('warn');
+  });
+});
+
+describe('workflowsApi.updateWorkflow', () => {
+  beforeEach(async () => {
+    const { callCoreRpc } = await import('../../coreRpcClient');
+    vi.mocked(callCoreRpc).mockReset();
+  });
+
+  it('forwards every optional field and rekeys allowedTools', async () => {
+    const { callCoreRpc } = await import('../../coreRpcClient');
+    vi.mocked(callCoreRpc).mockResolvedValueOnce({
+      workflow: { id: 'wf', name: 'WF', description: 'd', scope: 'user' as const },
+    });
+
+    await workflowsApi.updateWorkflow({
+      name: 'WF',
+      description: 'd',
+      whenToUse: 'when X happens',
+      scope: 'user',
+      license: 'MIT',
+      author: 'me',
+      tags: ['t'],
+      allowedTools: ['mcp/fs'],
+      inputs: [{ name: 'n', required: true }],
+    });
+
+    expect(callCoreRpc).toHaveBeenCalledWith({
+      method: 'openhuman.workflows_update',
+      params: {
+        name: 'WF',
+        description: 'd',
+        when_to_use: 'when X happens',
+        scope: 'user',
+        license: 'MIT',
+        author: 'me',
+        tags: ['t'],
+        'allowed-tools': ['mcp/fs'],
+        inputs: [{ name: 'n', required: true }],
+      },
+    });
+  });
+});
+
+describe('workflowsApi.listWorkflows', () => {
+  beforeEach(async () => {
+    const { callCoreRpc } = await import('../../coreRpcClient');
+    vi.mocked(callCoreRpc).mockReset();
+  });
+
+  it('reads the `workflows` result field', async () => {
+    const { callCoreRpc } = await import('../../coreRpcClient');
+    vi.mocked(callCoreRpc).mockResolvedValueOnce({
+      workflows: [
+        { id: 'a', name: 'A', description: '', scope: 'user' as const },
+        { id: 'b', name: 'B', description: '', scope: 'project' as const },
+      ],
+    });
+
+    const result = await workflowsApi.listWorkflows();
+
+    expect(callCoreRpc).toHaveBeenCalledWith({ method: 'openhuman.workflows_list' });
+    expect(result.map(w => w.id)).toEqual(['a', 'b']);
+  });
+
+  it('unwraps an envelope and defaults to [] when the field is absent', async () => {
+    const { callCoreRpc } = await import('../../coreRpcClient');
+    vi.mocked(callCoreRpc).mockResolvedValueOnce({ data: {} });
+    const result = await workflowsApi.listWorkflows();
+    expect(result).toEqual([]);
+  });
+
+  it('normalizes snake_case and legacy discovered workflow fields', async () => {
+    const { callCoreRpc } = await import('../../coreRpcClient');
+    vi.mocked(callCoreRpc).mockResolvedValueOnce({
+      workflows: [
         {
           id: 'hermes-demo',
           name: 'Hermes Demo',
@@ -49,7 +273,7 @@ describe('skillsApi.listSkills', () => {
       ],
     });
 
-    const result = await skillsApi.listSkills();
+    const result = await workflowsApi.listWorkflows();
 
     expect(callCoreRpc).toHaveBeenCalledWith({ method: 'openhuman.workflows_list' });
     expect(result[0].relatedSkills).toEqual(['browser-automation']);
@@ -59,158 +283,63 @@ describe('skillsApi.listSkills', () => {
   });
 });
 
-describe('skillsApi.createSkill', () => {
+describe('workflowsApi.readWorkflowResource', () => {
   beforeEach(async () => {
     const { callCoreRpc } = await import('../../coreRpcClient');
     vi.mocked(callCoreRpc).mockReset();
   });
 
-  it('forwards inputs to skills_create and rekeys allowedTools', async () => {
+  it('rekeys params to snake_case and normalizes the response', async () => {
     const { callCoreRpc } = await import('../../coreRpcClient');
     vi.mocked(callCoreRpc).mockResolvedValueOnce({
-      skill: {
-        id: 'my-skill',
-        name: 'my-skill',
-        description: 'does stuff',
-        version: '',
-        author: null,
-        tags: ['alpha'],
-        tools: ['mcp/fs'],
-        prompts: [],
-        location: '/home/u/.openhuman/skills/my-skill/SKILL.md',
-        resources: [],
-        scope: 'user',
-        legacy: false,
-        warnings: [],
-      },
+      workflow_id: 'wf',
+      relative_path: 'scripts/run.sh',
+      content: '#!/bin/sh\n',
+      bytes: 10,
     });
 
-    const result = await skillsApi.createSkill({
-      name: 'My Skill',
-      description: 'does stuff',
+    const result = await workflowsApi.readWorkflowResource({
+      workflowId: 'wf',
+      relativePath: 'scripts/run.sh',
+    });
+
+    expect(callCoreRpc).toHaveBeenCalledWith({
+      method: 'openhuman.workflows_read_resource',
+      params: { workflow_id: 'wf', relative_path: 'scripts/run.sh' },
+    });
+    expect(result).toEqual({
+      workflowId: 'wf',
+      relativePath: 'scripts/run.sh',
+      content: '#!/bin/sh\n',
+      bytes: 10,
+    });
+  });
+});
+
+describe('workflowsApi.uninstallWorkflow', () => {
+  beforeEach(async () => {
+    const { callCoreRpc } = await import('../../coreRpcClient');
+    vi.mocked(callCoreRpc).mockReset();
+  });
+
+  it('forwards the name and normalizes removed_path → removedPath', async () => {
+    const { callCoreRpc } = await import('../../coreRpcClient');
+    vi.mocked(callCoreRpc).mockResolvedValueOnce({
+      name: 'weather-helper',
+      removed_path: '/home/u/.openhuman/skills/weather-helper',
       scope: 'user',
-      tags: ['alpha'],
-      allowedTools: ['mcp/fs'],
     });
 
+    const result = await workflowsApi.uninstallWorkflow('weather-helper');
+
     expect(callCoreRpc).toHaveBeenCalledWith({
-      method: 'openhuman.workflows_create',
-      params: {
-        name: 'My Skill',
-        description: 'does stuff',
-        scope: 'user',
-        tags: ['alpha'],
-        'allowed-tools': ['mcp/fs'],
-      },
+      method: 'openhuman.workflows_uninstall',
+      params: { name: 'weather-helper' },
     });
-    expect(result.id).toBe('my-skill');
-    expect(result.scope).toBe('user');
-  });
-
-  it('omits optional fields when not provided', async () => {
-    const { callCoreRpc } = await import('../../coreRpcClient');
-    vi.mocked(callCoreRpc).mockResolvedValueOnce({
-      skill: {
-        id: 'minimal',
-        name: 'minimal',
-        description: 'd',
-        version: '',
-        author: null,
-        tags: [],
-        tools: [],
-        prompts: [],
-        location: null,
-        resources: [],
-        scope: 'user',
-        legacy: false,
-        warnings: [],
-      },
+    expect(result).toEqual({
+      name: 'weather-helper',
+      removedPath: '/home/u/.openhuman/skills/weather-helper',
+      scope: 'user',
     });
-
-    await skillsApi.createSkill({ name: 'minimal', description: 'd' });
-
-    const call = vi.mocked(callCoreRpc).mock.calls[0][0];
-    expect(call.params).toEqual({ name: 'minimal', description: 'd' });
-  });
-
-  it('unwraps an envelope response', async () => {
-    const { callCoreRpc } = await import('../../coreRpcClient');
-    vi.mocked(callCoreRpc).mockResolvedValueOnce({
-      data: {
-        skill: {
-          id: 'env',
-          name: 'env',
-          description: 'e',
-          version: '',
-          author: null,
-          tags: [],
-          tools: [],
-          prompts: [],
-          location: null,
-          resources: [],
-          scope: 'project',
-          legacy: false,
-          warnings: [],
-        },
-      },
-    });
-    const result = await skillsApi.createSkill({ name: 'env', description: 'e' });
-    expect(result.id).toBe('env');
-    expect(result.scope).toBe('project');
-  });
-});
-
-describe('skillsApi.installSkillFromUrl', () => {
-  beforeEach(async () => {
-    const { callCoreRpc } = await import('../../coreRpcClient');
-    vi.mocked(callCoreRpc).mockReset();
-  });
-
-  it('forwards url and rekeys timeoutSecs to timeout_secs', async () => {
-    const { callCoreRpc } = await import('../../coreRpcClient');
-    vi.mocked(callCoreRpc).mockResolvedValueOnce({
-      url: 'https://example.com/my-skill.tgz',
-      stdout: 'added my-skill',
-      stderr: '',
-      new_skills: ['my-skill'],
-    });
-
-    const result = await skillsApi.installSkillFromUrl({
-      url: 'https://example.com/my-skill.tgz',
-      timeoutSecs: 120,
-    });
-
-    expect(callCoreRpc).toHaveBeenCalledWith({
-      method: 'openhuman.workflows_install_from_url',
-      params: { url: 'https://example.com/my-skill.tgz', timeout_secs: 120 },
-    });
-    expect(result.newSkills).toEqual(['my-skill']);
-    expect(result.stdout).toBe('added my-skill');
-  });
-
-  it('omits timeout_secs when not provided and normalizes missing new_skills', async () => {
-    const { callCoreRpc } = await import('../../coreRpcClient');
-    vi.mocked(callCoreRpc).mockResolvedValueOnce({
-      url: 'https://example.com/x',
-      stdout: '',
-      stderr: '',
-      new_skills: undefined,
-    });
-
-    const result = await skillsApi.installSkillFromUrl({ url: 'https://example.com/x' });
-
-    const call = vi.mocked(callCoreRpc).mock.calls[0][0];
-    expect(call.params).toEqual({ url: 'https://example.com/x' });
-    expect(result.newSkills).toEqual([]);
-  });
-
-  it('unwraps an envelope response', async () => {
-    const { callCoreRpc } = await import('../../coreRpcClient');
-    vi.mocked(callCoreRpc).mockResolvedValueOnce({
-      data: { url: 'https://example.com/y', stdout: 'ok', stderr: 'warn', new_skills: ['y-skill'] },
-    });
-    const result = await skillsApi.installSkillFromUrl({ url: 'https://example.com/y' });
-    expect(result.newSkills).toEqual(['y-skill']);
-    expect(result.stderr).toBe('warn');
   });
 });

app/src/services/api/workflowsApi.test.ts

@@ -1,21 +1,21 @@
 import { beforeEach, describe, expect, it, vi } from 'vitest';
 
-import { skillsApi } from './skillsApi';
+import { workflowsApi } from './workflowsApi';
 
 const mockCallCoreRpc = vi.fn();
 vi.mock('../coreRpcClient', () => ({ callCoreRpc: (...a: unknown[]) => mockCallCoreRpc(...a) }));
 
-describe('skillsApi', () => {
+describe('workflowsApi', () => {
   beforeEach(() => {
     mockCallCoreRpc.mockReset();
   });
 
-  describe('createSkill', () => {
+  describe('createWorkflow', () => {
     it('includes inputs in params when non-empty', async () => {
       mockCallCoreRpc.mockResolvedValue({
-        skill: { id: 's', name: 'S', description: '', scope: 'user' as const },
+        workflow: { id: 's', name: 'S', description: '', scope: 'user' as const },
       });
-      await skillsApi.createSkill({
+      await workflowsApi.createWorkflow({
         name: 'S',
         description: 'desc',
         inputs: [{ name: 'repo', type: 'string' as const, description: 'repo', required: true }],
@@ -26,7 +26,7 @@ describe('skillsApi', () => {
     });
   });
 
-  describe('describeSkill', () => {
+  describe('describeWorkflow', () => {
     it('calls openhuman.workflows_describe with workflow_id', async () => {
       mockCallCoreRpc.mockResolvedValue({
         id: 'dev-workflow',
@@ -34,7 +34,7 @@ describe('skillsApi', () => {
         description: 'Auto dev',
         inputs: [],
       });
-      const result = await skillsApi.describeSkill('dev-workflow');
+      const result = await workflowsApi.describeWorkflow('dev-workflow');
       expect(mockCallCoreRpc).toHaveBeenCalledWith(
         expect.objectContaining({
           method: 'openhuman.workflows_describe',
@@ -48,15 +48,15 @@ describe('skillsApi', () => {
       mockCallCoreRpc.mockResolvedValue({
         data: { id: 'x', name: 'X', description: '', inputs: [], workflow_id: 'x' },
       });
-      const result = await skillsApi.describeSkill('x');
+      const result = await workflowsApi.describeWorkflow('x');
       expect(result.id).toBe('x');
     });
   });
 
-  describe('runSkill', () => {
+  describe('runWorkflow', () => {
     it('calls openhuman.workflows_run with workflow_id and inputs', async () => {
       mockCallCoreRpc.mockResolvedValue({ run_id: 'run-1', workflow_id: 's', log: '/tmp/log' });
-      const result = await skillsApi.runSkill('s', { repo: 'owner/repo' });
+      const result = await workflowsApi.runWorkflow('s', { repo: 'owner/repo' });
       expect(mockCallCoreRpc).toHaveBeenCalledWith(
         expect.objectContaining({
           method: 'openhuman.workflows_run',
@@ -68,7 +68,7 @@ describe('skillsApi', () => {
   });
 
   describe('readRunLog', () => {
-    it('calls skills_read_run_log with run_id', async () => {
+    it('calls workflows_read_run_log with run_id', async () => {
       mockCallCoreRpc.mockResolvedValue({
         bytes_read: 100,
         eof: false,
@@ -76,7 +76,7 @@ describe('skillsApi', () => {
         content: 'log line',
         offset: 100,
       });
-      const result = await skillsApi.readRunLog('run-1');
+      const result = await workflowsApi.readRunLog('run-1');
       expect(mockCallCoreRpc).toHaveBeenCalledWith(
         expect.objectContaining({
           method: 'openhuman.workflows_read_run_log',
@@ -94,7 +94,7 @@ describe('skillsApi', () => {
         content: '',
         offset: 500,
       });
-      await skillsApi.readRunLog('run-2', 200, 4096);
+      await workflowsApi.readRunLog('run-2', 200, 4096);
       expect(mockCallCoreRpc).toHaveBeenCalledWith(
         expect.objectContaining({
           params: expect.objectContaining({ run_id: 'run-2', offset: 200, max_bytes: 4096 }),
@@ -106,13 +106,13 @@ describe('skillsApi', () => {
   describe('recentRuns', () => {
     it('returns scanned runs array', async () => {
       mockCallCoreRpc.mockResolvedValue({ runs: [] });
-      const result = await skillsApi.recentRuns();
+      const result = await workflowsApi.recentRuns();
       expect(Array.isArray(result)).toBe(true);
     });
 
     it('passes workflow_id filter when provided', async () => {
       mockCallCoreRpc.mockResolvedValue({ runs: [] });
-      await skillsApi.recentRuns('dev-workflow', 5);
+      await workflowsApi.recentRuns('dev-workflow', 5);
       expect(mockCallCoreRpc).toHaveBeenCalledWith(
         expect.objectContaining({
           params: expect.objectContaining({ workflow_id: 'dev-workflow', limit: 5 }),
@@ -121,12 +121,12 @@ describe('skillsApi', () => {
     });
   });
 
-  describe('createSkill (optional fields)', () => {
+  describe('createWorkflow (optional fields)', () => {
     it('forwards when_to_use, scope, license, author, tags, allowed-tools', async () => {
       mockCallCoreRpc.mockResolvedValue({
-        skill: { id: 's', name: 'S', description: '', scope: 'user' as const },
+        workflow: { id: 's', name: 'S', description: '', scope: 'user' as const },
       });
-      await skillsApi.createSkill({
+      await workflowsApi.createWorkflow({
         name: 'S',
         description: 'desc',
         whenToUse: 'when asked',
@@ -153,20 +153,20 @@ describe('skillsApi', () => {
 
     it('omits when_to_use when blank', async () => {
       mockCallCoreRpc.mockResolvedValue({
-        skill: { id: 's', name: 'S', description: '', scope: 'user' as const },
+        workflow: { id: 's', name: 'S', description: '', scope: 'user' as const },
       });
-      await skillsApi.createSkill({ name: 'S', description: 'd', whenToUse: '   ' });
+      await workflowsApi.createWorkflow({ name: 'S', description: 'd', whenToUse: '   ' });
       const params = mockCallCoreRpc.mock.calls[0][0].params;
       expect(params).not.toHaveProperty('when_to_use');
     });
   });
 
-  describe('updateSkill', () => {
+  describe('updateWorkflow', () => {
     it('calls openhuman.workflows_update and returns the skill', async () => {
       mockCallCoreRpc.mockResolvedValue({
-        skill: { id: 'wf', name: 'WF', description: 'd', scope: 'user' as const },
+        workflow: { id: 'wf', name: 'WF', description: 'd', scope: 'user' as const },
       });
-      const result = await skillsApi.updateSkill({
+      const result = await workflowsApi.updateWorkflow({
         name: 'WF',
         description: 'd',
         whenToUse: 'edit trigger',
@@ -187,9 +187,9 @@ describe('skillsApi', () => {
 
     it('unwraps the data-envelope shape', async () => {
       mockCallCoreRpc.mockResolvedValue({
-        data: { skill: { id: 'wf2', name: 'WF2', description: '', scope: 'user' as const } },
+        data: { workflow: { id: 'wf2', name: 'WF2', description: '', scope: 'user' as const } },
       });
-      const result = await skillsApi.updateSkill({ name: 'WF2', description: 'd' });
+      const result = await workflowsApi.updateWorkflow({ name: 'WF2', description: 'd' });
       expect(result.id).toBe('wf2');
     });
   });
@@ -197,7 +197,7 @@ describe('skillsApi', () => {
   describe('cancelRun', () => {
     it('calls openhuman.workflows_cancel with run_id and returns cancelled', async () => {
       mockCallCoreRpc.mockResolvedValue({ cancelled: true });
-      const result = await skillsApi.cancelRun('run-9');
+      const result = await workflowsApi.cancelRun('run-9');
       expect(mockCallCoreRpc).toHaveBeenCalledWith(
         expect.objectContaining({
           method: 'openhuman.workflows_cancel',
@@ -209,7 +209,7 @@ describe('skillsApi', () => {
 
     it('returns false when the run was not live (envelope shape)', async () => {
       mockCallCoreRpc.mockResolvedValue({ data: { cancelled: false } });
-      const result = await skillsApi.cancelRun('gone');
+      const result = await workflowsApi.cancelRun('gone');
       expect(result).toBe(false);
     });
   });

app/src/services/api/workflowsApi.ts

@@ -2,15 +2,15 @@ import debug from 'debug';
 
 import { callCoreRpc } from '../coreRpcClient';
 
-const log = debug('skillsApi');
+const log = debug('workflowsApi');
 
 /**
  * Scope a skill was discovered in.
  *
- * Mirrors `openhuman::skills::ops::SkillScope` on the Rust side — serialized
+ * Mirrors `openhuman::skills::ops::WorkflowScope` on the Rust side — serialized
  * as a lowercase string (`"user" | "project" | "legacy"`).
  */
-export type SkillScope = 'user' | 'project' | 'legacy';
+export type WorkflowScope = 'user' | 'project' | 'legacy';
 
 /**
  * Wire-format representation of a discovered skill returned by
@@ -19,7 +19,7 @@ export type SkillScope = 'user' | 'project' | 'legacy';
  * Paths are intentionally serialized as strings (not URLs) to avoid lossy
  * conversions on non-UTF-8 filesystems.
  */
-export interface SkillSummary {
+export interface WorkflowSummary {
   /** Stable identifier — equal to `name` on the Rust side. */
   id: string;
   /** Display name, from frontmatter or directory. */
@@ -47,18 +47,18 @@ export interface SkillSummary {
   /** Bundled resource files, relative to the skill root. */
   resources: string[];
   /** Where the skill came from. */
-  scope: SkillScope;
+  scope: WorkflowScope;
   /** True when loaded from the legacy `skills/` layout. */
   legacy: boolean;
   /** Non-fatal parse warnings to surface in the UI. */
   warnings: string[];
 }
 
-interface SkillsListResult {
-  skills: RawSkillSummary[];
+interface WorkflowsListResult {
+  workflows: RawWorkflowSummary[];
 }
 
-type RawSkillSummary = Omit<SkillSummary, 'platforms' | 'relatedSkills' | 'sourceFormat'> & {
+type RawWorkflowSummary = Omit<WorkflowSummary, 'platforms' | 'relatedSkills' | 'sourceFormat'> & {
   platforms?: string[];
   related_skills?: string[];
   relatedSkills?: string[];
@@ -69,9 +69,9 @@ type RawSkillSummary = Omit<SkillSummary, 'platforms' | 'relatedSkills' | 'sourc
 /**
  * Result of `openhuman.workflows_read_resource`.
  */
-export interface SkillResourceContent {
+export interface WorkflowResourceContent {
   /** Echo of the requested skill id. */
-  skillId: string;
+  workflowId: string;
   /** Echo of the requested relative path. */
   relativePath: string;
   /** UTF-8 file contents (<= 128 KB). */
@@ -80,7 +80,7 @@ export interface SkillResourceContent {
   bytes: number;
 }
 
-interface RawSkillsReadResourceResult {
+interface RawWorkflowsReadResourceResult {
   workflow_id: string;
   relative_path: string;
   content: string;
@@ -102,14 +102,14 @@ interface RawSkillsReadResourceResult {
  * to `true` on the Rust side when omitted (we send it explicitly to
  * stay loud).
  */
-export interface CreateSkillInputDef {
+export interface CreateWorkflowInputDef {
   name: string;
   description?: string;
   required: boolean;
   type?: 'string' | 'integer' | 'boolean';
 }
 
-export interface CreateSkillInput {
+export interface CreateWorkflowInput {
   name: string;
   description: string;
   /**
@@ -119,7 +119,7 @@ export interface CreateSkillInput {
    * `skill.toml` `when_to_use`; falls back to `description` when omitted.
    */
   whenToUse?: string;
-  scope?: SkillScope;
+  scope?: WorkflowScope;
   license?: string;
   author?: string;
   tags?: string[];
@@ -130,11 +130,11 @@ export interface CreateSkillInput {
    * the Skills Runner can render dynamic form controls per input.
    * Omit / pass `[]` to scaffold an input-less skill.
    */
-  inputs?: CreateSkillInputDef[];
+  inputs?: CreateWorkflowInputDef[];
 }
 
-interface RawSkillsCreateResult {
-  skill: RawSkillSummary;
+interface RawWorkflowsCreateResult {
+  workflow: RawWorkflowSummary;
 }
 
 /**
@@ -143,7 +143,7 @@ interface RawSkillsCreateResult {
  * `timeoutSecs` is optional — the Rust side defaults to 60s and caps at
  * 600s. Values outside that range are clamped server-side.
  */
-export interface InstallSkillFromUrlInput {
+export interface InstallWorkflowFromUrlInput {
   url: string;
   timeoutSecs?: number;
 }
@@ -151,24 +151,24 @@ export interface InstallSkillFromUrlInput {
 /**
  * Result of `openhuman.workflows_install_from_url`.
  *
- * `newSkills` lists skill ids that appeared post-install (diff vs the
+ * `newWorkflows` lists skill ids that appeared post-install (diff vs the
  * pre-install snapshot). `stdout` holds a human-readable diagnostic summary
  * (bytes fetched, target path); `stderr` holds non-fatal frontmatter parse
  * warnings joined by newlines. There is no subprocess — the Rust side fetches
  * SKILL.md directly over HTTPS.
  */
-export interface InstallSkillFromUrlResult {
+export interface InstallWorkflowFromUrlResult {
   url: string;
   stdout: string;
   stderr: string;
-  newSkills: string[];
+  newWorkflows: string[];
 }
 
-interface RawInstallSkillFromUrlResult {
+interface RawInstallWorkflowFromUrlResult {
   url: string;
   stdout: string;
   stderr: string;
-  new_skills: string[];
+  new_workflows: string[];
 }
 
 /**
@@ -178,16 +178,16 @@ interface RawInstallSkillFromUrlResult {
  * canonicalised on-disk path that was deleted — surface it in success toasts
  * so the user can confirm exactly what was removed.
  */
-export interface UninstallSkillResult {
+export interface UninstallWorkflowResult {
   name: string;
   removedPath: string;
-  scope: SkillScope;
+  scope: WorkflowScope;
 }
 
-interface RawUninstallSkillResult {
+interface RawUninstallWorkflowResult {
   name: string;
   removed_path: string;
-  scope: SkillScope;
+  scope: WorkflowScope;
 }
 
 interface Envelope<T> {
@@ -204,7 +204,7 @@ function unwrapEnvelope<T>(response: Envelope<T> | T): T {
   return response as T;
 }
 
-function normalizeSkillSummary(raw: RawSkillSummary): SkillSummary {
+function normalizeWorkflowSummary(raw: RawWorkflowSummary): WorkflowSummary {
   return {
     ...raw,
     platforms: raw.platforms ?? [],
@@ -213,17 +213,17 @@ function normalizeSkillSummary(raw: RawSkillSummary): SkillSummary {
   };
 }
 
-export const skillsApi = {
+export const workflowsApi = {
   /** Enumerate SKILL.md / legacy skills visible in the active workspace. */
-  listSkills: async (): Promise<SkillSummary[]> => {
-    log('listSkills: request');
-    const response = await callCoreRpc<Envelope<SkillsListResult> | SkillsListResult>({
+  listWorkflows: async (): Promise<WorkflowSummary[]> => {
+    log('listWorkflows: request');
+    const response = await callCoreRpc<Envelope<WorkflowsListResult> | WorkflowsListResult>({
       method: 'openhuman.workflows_list',
     });
     const result = unwrapEnvelope(response);
-    const skills = (result?.skills ?? []).map(normalizeSkillSummary);
-    log('listSkills: response count=%d', skills.length);
-    return skills;
+    const workflows = (result?.workflows ?? []).map(normalizeWorkflowSummary);
+    log('listWorkflows: response count=%d', workflows.length);
+    return workflows;
   },
 
   /**
@@ -231,28 +231,28 @@ export const skillsApi = {
    * traversal, symlink escape, non-UTF-8 payloads, or files larger than
    * 128 KB — the caller surfaces the error string verbatim in the drawer.
    */
-  readSkillResource: async ({
-    skillId,
+  readWorkflowResource: async ({
+    workflowId,
     relativePath,
   }: {
-    skillId: string;
+    workflowId: string;
     relativePath: string;
-  }): Promise<SkillResourceContent> => {
-    log('readSkillResource: request skillId=%s path=%s', skillId, relativePath);
+  }): Promise<WorkflowResourceContent> => {
+    log('readWorkflowResource: request workflowId=%s path=%s', workflowId, relativePath);
     const response = await callCoreRpc<
-      Envelope<RawSkillsReadResourceResult> | RawSkillsReadResourceResult
+      Envelope<RawWorkflowsReadResourceResult> | RawWorkflowsReadResourceResult
     >({
       method: 'openhuman.workflows_read_resource',
-      params: { workflow_id: skillId, relative_path: relativePath },
+      params: { workflow_id: workflowId, relative_path: relativePath },
     });
     const raw = unwrapEnvelope(response);
-    const normalized: SkillResourceContent = {
-      skillId: raw.workflow_id,
+    const normalized: WorkflowResourceContent = {
+      workflowId: raw.workflow_id,
       relativePath: raw.relative_path,
       content: raw.content,
       bytes: raw.bytes,
     };
-    log('readSkillResource: response bytes=%d', normalized.bytes);
+    log('readWorkflowResource: response bytes=%d', normalized.bytes);
     return normalized;
   },
 
@@ -260,12 +260,14 @@ export const skillsApi = {
    * Scaffold a new SKILL.md skill via `openhuman.workflows_create`.
    *
    * The Rust side slugifies the name, writes `SKILL.md` with the supplied
-   * frontmatter, and returns the freshly-discovered `SkillSummary` so the
+   * frontmatter, and returns the freshly-discovered `WorkflowSummary` so the
    * caller can insert the new row into the grid without a full refetch.
    */
-  createSkill: async (input: CreateSkillInput): Promise<SkillSummary> => {
-    log('createSkill: request name=%s scope=%s', input.name, input.scope ?? 'default');
-    const response = await callCoreRpc<Envelope<RawSkillsCreateResult> | RawSkillsCreateResult>({
+  createWorkflow: async (input: CreateWorkflowInput): Promise<WorkflowSummary> => {
+    log('createWorkflow: request name=%s scope=%s', input.name, input.scope ?? 'default');
+    const response = await callCoreRpc<
+      Envelope<RawWorkflowsCreateResult> | RawWorkflowsCreateResult
+    >({
       method: 'openhuman.workflows_create',
       params: {
         name: input.name,
@@ -282,9 +284,9 @@ export const skillsApi = {
       },
     });
     const raw = unwrapEnvelope(response);
-    const skill = normalizeSkillSummary(raw.skill);
-    log('createSkill: response id=%s', skill.id);
-    return skill;
+    const workflow = normalizeWorkflowSummary(raw.workflow);
+    log('createWorkflow: response id=%s', workflow.id);
+    return workflow;
   },
 
   /**
@@ -293,9 +295,11 @@ export const skillsApi = {
    * slug — rewriting frontmatter + workflow.toml while preserving the
    * hand-authored SKILL.md/WORKFLOW.md body.
    */
-  updateSkill: async (input: CreateSkillInput): Promise<SkillSummary> => {
-    log('updateSkill: request name=%s scope=%s', input.name, input.scope ?? 'default');
-    const response = await callCoreRpc<Envelope<RawSkillsCreateResult> | RawSkillsCreateResult>({
+  updateWorkflow: async (input: CreateWorkflowInput): Promise<WorkflowSummary> => {
+    log('updateWorkflow: request name=%s scope=%s', input.name, input.scope ?? 'default');
+    const response = await callCoreRpc<
+      Envelope<RawWorkflowsCreateResult> | RawWorkflowsCreateResult
+    >({
       method: 'openhuman.workflows_update',
       params: {
         name: input.name,
@@ -312,9 +316,9 @@ export const skillsApi = {
       },
     });
     const raw = unwrapEnvelope(response);
-    const skill = normalizeSkillSummary(raw.skill);
-    log('updateSkill: response id=%s', skill.id);
-    return skill;
+    const workflow = normalizeWorkflowSummary(raw.workflow);
+    log('updateWorkflow: response id=%s', workflow.id);
+    return workflow;
   },
 
   /**
@@ -327,12 +331,12 @@ export const skillsApi = {
    * is normalised to its `raw.githubusercontent.com` equivalent. Size is
    * capped at 1 MiB; timeout default 60s, max 600s.
    */
-  installSkillFromUrl: async (
-    input: InstallSkillFromUrlInput
-  ): Promise<InstallSkillFromUrlResult> => {
-    log('installSkillFromUrl: request url=%s', input.url);
+  installWorkflowFromUrl: async (
+    input: InstallWorkflowFromUrlInput
+  ): Promise<InstallWorkflowFromUrlResult> => {
+    log('installWorkflowFromUrl: request url=%s', input.url);
     const response = await callCoreRpc<
-      Envelope<RawInstallSkillFromUrlResult> | RawInstallSkillFromUrlResult
+      Envelope<RawInstallWorkflowFromUrlResult> | RawInstallWorkflowFromUrlResult
     >({
       method: 'openhuman.workflows_install_from_url',
       params: {
@@ -341,15 +345,15 @@ export const skillsApi = {
       },
     });
     const raw = unwrapEnvelope(response);
-    const normalized: InstallSkillFromUrlResult = {
+    const normalized: InstallWorkflowFromUrlResult = {
       url: raw.url,
       stdout: raw.stdout,
       stderr: raw.stderr,
-      newSkills: raw.new_skills ?? [],
+      newWorkflows: raw.new_workflows ?? [],
     };
     log(
-      'installSkillFromUrl: response new=%d stdout=%d stderr=%d',
-      normalized.newSkills.length,
+      'installWorkflowFromUrl: response new=%d stdout=%d stderr=%d',
+      normalized.newWorkflows.length,
       normalized.stdout.length,
       normalized.stderr.length
     );
@@ -365,37 +369,41 @@ export const skillsApi = {
    * canonicalises paths and refuses names with separators / traversal
    * sequences / anything outside the skills root.
    */
-  uninstallSkill: async (name: string): Promise<UninstallSkillResult> => {
-    log('uninstallSkill: request name=%s', name);
-    const response = await callCoreRpc<Envelope<RawUninstallSkillResult> | RawUninstallSkillResult>(
-      { method: 'openhuman.workflows_uninstall', params: { name } }
-    );
+  uninstallWorkflow: async (name: string): Promise<UninstallWorkflowResult> => {
+    log('uninstallWorkflow: request name=%s', name);
+    const response = await callCoreRpc<
+      Envelope<RawUninstallWorkflowResult> | RawUninstallWorkflowResult
+    >({ method: 'openhuman.workflows_uninstall', params: { name } });
     const raw = unwrapEnvelope(response);
-    const normalized: UninstallSkillResult = {
+    const normalized: UninstallWorkflowResult = {
       name: raw.name,
       removedPath: raw.removed_path,
       scope: raw.scope,
     };
-    log('uninstallSkill: response name=%s removedPath=%s', normalized.name, normalized.removedPath);
+    log(
+      'uninstallWorkflow: response name=%s removedPath=%s',
+      normalized.name,
+      normalized.removedPath
+    );
     return normalized;
   },
 
   /**
    * Fetch the declared `[[inputs]]` for a single skill plus its display
-   * metadata. Lightweight companion to `listSkills` — `SkillSummary` rows
+   * metadata. Lightweight companion to `listWorkflows` — `WorkflowSummary` rows
    * (used by the catalog grid) deliberately don't include input
    * declarations, so the Skills Runner panel calls this once when the
    * user picks a skill from the dropdown so it can render the right form
    * controls.
    */
-  describeSkill: async (skillId: string): Promise<SkillDescription> => {
-    log('describeSkill: request skillId=%s', skillId);
-    const response = await callCoreRpc<Envelope<SkillDescription> | SkillDescription>({
+  describeWorkflow: async (workflowId: string): Promise<WorkflowDescription> => {
+    log('describeWorkflow: request workflowId=%s', workflowId);
+    const response = await callCoreRpc<Envelope<WorkflowDescription> | WorkflowDescription>({
       method: 'openhuman.workflows_describe',
-      params: { workflow_id: skillId },
+      params: { workflow_id: workflowId },
     });
     const raw = unwrapEnvelope(response);
-    log('describeSkill: response inputs=%d', raw.inputs.length);
+    log('describeWorkflow: response inputs=%d', raw.inputs.length);
     return raw;
   },
 
@@ -406,14 +414,17 @@ export const skillsApi = {
    * autonomous work continues in the background and finishes with
    * status `DONE` / `DEGENERATE` / `FAILED` in the run log.
    */
-  runSkill: async (skillId: string, inputs: Record<string, unknown>): Promise<SkillRunStarted> => {
-    log('runSkill: request skillId=%s', skillId);
-    const response = await callCoreRpc<Envelope<SkillRunStarted> | SkillRunStarted>({
+  runWorkflow: async (
+    workflowId: string,
+    inputs: Record<string, unknown>
+  ): Promise<WorkflowRunStarted> => {
+    log('runWorkflow: request workflowId=%s', workflowId);
+    const response = await callCoreRpc<Envelope<WorkflowRunStarted> | WorkflowRunStarted>({
       method: 'openhuman.workflows_run',
-      params: { workflow_id: skillId, inputs },
+      params: { workflow_id: workflowId, inputs },
     });
     const raw = unwrapEnvelope(response);
-    log('runSkill: response runId=%s log=%s', raw.run_id, raw.log);
+    log('runWorkflow: response runId=%s log=%s', raw.run_id, raw.log);
     return raw;
   },
 
@@ -459,13 +470,13 @@ export const skillsApi = {
 
   /**
    * Recent autonomous skill runs from `<workspace>/skills/.runs/`. Sorted
-   * by start time descending. Pass `skillId` to filter to one skill,
+   * by start time descending. Pass `workflowId` to filter to one skill,
    * omit for cross-skill. `limit` defaults to 20 (max 100).
    */
-  recentRuns: async (skillId?: string, limit?: number): Promise<ScannedRun[]> => {
-    log('recentRuns: request skillId=%s limit=%s', skillId ?? '*', limit ?? 'default');
+  recentRuns: async (workflowId?: string, limit?: number): Promise<ScannedRun[]> => {
+    log('recentRuns: request workflowId=%s limit=%s', workflowId ?? '*', limit ?? 'default');
     const params: Record<string, unknown> = {};
-    if (skillId !== undefined) params.workflow_id = skillId;
+    if (workflowId !== undefined) params.workflow_id = workflowId;
     if (limit !== undefined) params.limit = limit;
     const response = await callCoreRpc<Envelope<{ runs: ScannedRun[] }> | { runs: ScannedRun[] }>({
       method: 'openhuman.workflows_recent_runs',
@@ -482,7 +493,7 @@ export const skillsApi = {
  * `openhuman.workflows_describe`. The FE renders one form control per entry:
  * `string`/`integer`/`boolean` map to text/number/checkbox controls.
  */
-export interface SkillInputDescription {
+export interface WorkflowInputDescription {
   name: string;
   description: string;
   required: boolean;
@@ -491,15 +502,15 @@ export interface SkillInputDescription {
 }
 
 /** Wire shape returned by `openhuman.workflows_describe`. */
-export interface SkillDescription {
+export interface WorkflowDescription {
   id: string;
   display_name: string;
   when_to_use: string;
-  inputs: SkillInputDescription[];
+  inputs: WorkflowInputDescription[];
 }
 
 /** Wire shape returned by `openhuman.workflows_run` (fire-and-forget). */
-export interface SkillRunStarted {
+export interface WorkflowRunStarted {
   run_id: string;
   status: string; // "started"
   workflow_id: string;

src/core/jsonrpc.rs

@@ -2116,6 +2116,16 @@ pub async fn bootstrap_core_runtime(host_kind: crate::core::types::HostKind) {
         action_dir,
     );
 
+    // --- Triggered-workflow subscriber ---
+    // Install on the always-run serve boot, not only inside `start_channels`
+    // (skipped for web-chat-only cores with no messaging integrations, and when
+    // `OPENHUMAN_DISABLE_CHANNEL_LISTENERS=1`). Without this, any workflow
+    // declaring `triggers:` was silently ignored on web-chat-only desktop
+    // installs. Idempotent — shares a process-global OnceLock with the
+    // `start_channels` site so it registers exactly once regardless of which
+    // path runs first. (Matching only for now; activation handoff still pending.)
+    crate::openhuman::workflows::bus::ensure_triggered_workflow_subscriber(&workspace_dir);
+
     // --- Approval gate (#1339) ---
     // ON by default; opt out with `OPENHUMAN_APPROVAL_GATE=0` (or `false`).
     // Prompt-class `external_effect()` tool calls route through

src/openhuman/agent/debug/mod.rs

@@ -418,7 +418,7 @@ async fn render_integrations_agent(config: &Config, toolkit: &str) -> Result<Dum
         model_name: &model_name,
         agent_id: INTEGRATIONS_AGENT_ID,
         tools: &prompt_tools,
-        skills: agent.skills(),
+        workflows: agent.workflows(),
         dispatcher_instructions: "",
         learned: LearnedContextData::default(),
         visible_tool_names: &empty_visible,

src/openhuman/agent/harness/fork_context.rs

@@ -70,9 +70,9 @@ pub struct ParentExecutionContext {
     /// dispatcher choice, …).
     pub agent_config: AgentConfig,
 
-    /// Skills loaded into the parent. Sub-agents that don't strip the
-    /// skills catalog inherit this list.
-    pub skills: Arc<Vec<Workflow>>,
+    /// Workflows loaded into the parent. Sub-agents that don't strip the
+    /// workflows catalog inherit this list.
+    pub workflows: Arc<Vec<Workflow>>,
 
     /// Memory context loaded for the current turn. Auto-injected into
     /// subagent prompts so they have access to conversation history and

src/openhuman/agent/harness/harness_gap_tests.rs

@@ -531,7 +531,7 @@ fn datetime_section_output_matches_iso8601_date_and_utc_offset_pattern() {
         model_name: "test-model",
         agent_id: "",
         tools: EMPTY_TOOLS,
-        skills: &[],
+        workflows: &[],
         dispatcher_instructions: "",
         learned: crate::openhuman::agent::prompts::LearnedContextData::default(),
         visible_tool_names: &EMPTY_FILTER,

src/openhuman/agent/harness/session/builder/factory.rs

@@ -1050,7 +1050,7 @@ impl Agent {
             .temperature(effective_temperature)
             .workspace_dir(config.workspace_dir.clone())
             .action_dir(config.action_dir.clone())
-            .skills(crate::openhuman::workflows::load_workflow_metadata(
+            .workflows(crate::openhuman::workflows::load_workflow_metadata(
                 &config.workspace_dir,
             ))
             .auto_save(config.memory.auto_save)

src/openhuman/agent/harness/session/builder/setters.rs

@@ -32,7 +32,7 @@ impl AgentBuilder {
             temperature: None,
             workspace_dir: None,
             action_dir: None,
-            skills: None,
+            workflows: None,
             auto_save: None,
             post_turn_hooks: Vec::new(),
             learning_enabled: false,
@@ -160,8 +160,8 @@ impl AgentBuilder {
     }
 
     /// Sets the skills available to the agent.
-    pub fn skills(mut self, skills: Vec<crate::openhuman::workflows::Workflow>) -> Self {
-        self.skills = Some(skills);
+    pub fn workflows(mut self, skills: Vec<crate::openhuman::workflows::Workflow>) -> Self {
+        self.workflows = Some(skills);
         self
     }
 
@@ -517,7 +517,7 @@ impl AgentBuilder {
             temperature: self.temperature.unwrap_or(0.7),
             workspace_dir,
             action_dir,
-            skills: self.skills.unwrap_or_default(),
+            workflows: self.workflows.unwrap_or_default(),
             auto_save: self.auto_save.unwrap_or(false),
             last_memory_context: None,
             last_turn_citations: Vec::new(),

src/openhuman/agent/harness/session/runtime.rs

@@ -109,9 +109,9 @@ impl Agent {
         self.temperature
     }
 
-    /// The agent's loaded skills, if any.
-    pub fn skills(&self) -> &[crate::openhuman::workflows::Workflow] {
-        &self.skills
+    /// The agent's loaded workflows, if any.
+    pub fn workflows(&self) -> &[crate::openhuman::workflows::Workflow] {
+        &self.workflows
     }
 
     /// Active Composio integrations fetched at session start.

src/openhuman/agent/harness/session/runtime_tests.rs

@@ -271,7 +271,7 @@ fn accessors_and_history_reset_expose_agent_runtime_state() {
     });
     let mut agent = make_agent(provider);
     agent.history = vec![ConversationMessage::Chat(ChatMessage::system("sys"))];
-    agent.skills = vec![crate::openhuman::workflows::Workflow {
+    agent.workflows = vec![crate::openhuman::workflows::Workflow {
         name: "demo".into(),
         ..Default::default()
     }];
@@ -283,7 +283,7 @@ fn accessors_and_history_reset_expose_agent_runtime_state() {
     assert_eq!(agent.workspace_dir(), agent.workspace_dir.as_path());
     assert_eq!(agent.model_name(), agent.model_name);
     assert_eq!(agent.temperature(), agent.temperature);
-    assert_eq!(agent.skills().len(), 1);
+    assert_eq!(agent.workflows().len(), 1);
     assert_eq!(
         agent.agent_config().max_tool_iterations,
         agent.config.max_tool_iterations

src/openhuman/agent/harness/session/turn/context.rs

@@ -293,7 +293,7 @@ impl Agent {
             model_name: &self.model_name,
             agent_id: &self.agent_definition_name,
             tools: &prompt_tools,
-            skills: &self.skills,
+            workflows: &self.workflows,
             dispatcher_instructions: &instructions,
             learned,
             visible_tool_names: &prompt_visible_tool_names,

src/openhuman/agent/harness/session/turn/core.rs

@@ -273,11 +273,11 @@ impl Agent {
         // heuristic and size cap rationale.
         let enriched = {
             use crate::openhuman::workflows::inject;
-            let matches = inject::match_workflows(&self.skills, user_message);
+            let matches = inject::match_workflows(&self.workflows, user_message);
             if matches.is_empty() {
                 log::debug!(
-                    "[skills:inject] no skill matches for user message (skill_catalog_len={})",
-                    self.skills.len()
+                    "[workflows:inject] no skill matches for user message (skill_catalog_len={})",
+                    self.workflows.len()
                 );
                 enriched
             } else {
@@ -288,7 +288,7 @@ impl Agent {
                 );
                 let matched_count = injection.decisions.iter().filter(|d| d.matched).count();
                 log::info!(
-                    "[skills:inject] summary candidates={} matched={} injected_bytes={} truncated_any={}",
+                    "[workflows:inject] summary candidates={} matched={} injected_bytes={} truncated_any={}",
                     injection.decisions.len(),
                     matched_count,
                     injection.injected_bytes,

src/openhuman/agent/harness/session/turn/tools.rs

@@ -124,7 +124,7 @@ impl Agent {
             workspace_dir: self.workspace_dir.clone(),
             memory: Arc::clone(&self.memory),
             agent_config: self.config.clone(),
-            skills: Arc::new(self.skills.clone()),
+            workflows: Arc::new(self.workflows.clone()),
             memory_context: Arc::new(self.last_memory_context.clone()),
             session_id: self.event_session_id().to_string(),
             channel: self.event_channel().to_string(),

src/openhuman/agent/harness/session/turn_tests.rs

@@ -447,7 +447,7 @@ fn trim_history_snaps_past_orphaned_tool_results() {
 fn build_parent_context_and_sanitize_helpers_cover_snapshot_paths() {
     let mut agent = make_agent(None);
     agent.last_memory_context = Some("remember this".into());
-    agent.skills = vec![crate::openhuman::workflows::Workflow {
+    agent.workflows = vec![crate::openhuman::workflows::Workflow {
         name: "demo".into(),
         ..Default::default()
     }];
@@ -458,7 +458,7 @@ fn build_parent_context_and_sanitize_helpers_cover_snapshot_paths() {
     assert_eq!(parent.memory_context.as_deref(), Some("remember this"));
     assert_eq!(parent.session_id, "turn-test-session");
     assert_eq!(parent.channel, "turn-test-channel");
-    assert_eq!(parent.skills.len(), 1);
+    assert_eq!(parent.workflows.len(), 1);
 
     assert_eq!(sanitize_learned_entry("   "), "");
     assert_eq!(

src/openhuman/agent/harness/session/types.rs

@@ -56,7 +56,7 @@ pub struct Agent {
     pub(super) temperature: f64,
     pub(super) workspace_dir: std::path::PathBuf,
     pub(super) action_dir: std::path::PathBuf,
-    pub(super) skills: Vec<crate::openhuman::workflows::Workflow>,
+    pub(super) workflows: Vec<crate::openhuman::workflows::Workflow>,
     /// Agent workflows discovered at session start.
     pub(super) auto_save: bool,
     /// Last memory context loaded for the current turn. Stored so it can
@@ -275,7 +275,7 @@ pub struct AgentBuilder {
     pub(super) temperature: Option<f64>,
     pub(super) workspace_dir: Option<std::path::PathBuf>,
     pub(super) action_dir: Option<std::path::PathBuf>,
-    pub(super) skills: Option<Vec<crate::openhuman::workflows::Workflow>>,
+    pub(super) workflows: Option<Vec<crate::openhuman::workflows::Workflow>>,
     /// Agent workflows to surface in the prompt. Populated from `load_workflows`
     /// at session start; defaults to empty when not explicitly set.
     pub(super) auto_save: Option<bool>,

src/openhuman/agent/harness/subagent_runner/ops/runner.rs

@@ -563,7 +563,7 @@ async fn run_typed_mode(
         model_name: &model,
         agent_id: &definition.id,
         tools: &prompt_tools,
-        skills: &parent.skills,
+        workflows: &parent.workflows,
         dispatcher_instructions: &dispatcher_instructions,
         learned: crate::openhuman::context::prompt::LearnedContextData::default(),
         visible_tool_names: &visible_tool_names,

src/openhuman/agent/harness/subagent_runner/ops_tests.rs

@@ -326,7 +326,7 @@ fn make_parent(provider: Arc<dyn Provider>, tools: Vec<Box<dyn Tool>>) -> Parent
         workspace_dir: std::env::temp_dir(),
         memory: noop_memory(),
         agent_config: crate::openhuman::config::AgentConfig::default(),
-        skills: Arc::new(vec![]),
+        workflows: Arc::new(vec![]),
         memory_context: Arc::new(None),
         session_id: "test-session".into(),
         channel: "test".into(),

src/openhuman/agent/harness/test_support_tests.rs

@@ -1599,7 +1599,7 @@ async fn orchestrator_prompt_drives_composio_call_via_delegation_chain() {
             model_name: "test",
             agent_id: "orchestrator",
             tools: &[],
-            skills: &[],
+            workflows: &[],
             dispatcher_instructions: "",
             learned: LearnedContextData::default(),
             visible_tool_names: EMPTY.get_or_init(HashSet::new),

src/openhuman/agent/profiles.rs

@@ -870,7 +870,7 @@ mod tests {
             model_name: "test-model",
             agent_id: "orchestrator",
             tools: &[],
-            skills: &[],
+            workflows: &[],
             dispatcher_instructions: "",
             learned: LearnedContextData::default(),
             visible_tool_names: &visible_tool_names,

src/openhuman/agent/prompts/mod_tests.rs

@@ -58,7 +58,7 @@ fn prompt_builder_assembles_sections() {
         model_name: "test-model",
         agent_id: "",
         tools: &prompt_tools,
-        skills: &[],
+        workflows: &[],
         dispatcher_instructions: "instr",
         learned: LearnedContextData::default(),
         visible_tool_names: &NO_FILTER,
@@ -92,7 +92,7 @@ fn identity_section_creates_missing_workspace_files() {
         model_name: "test-model",
         agent_id: "",
         tools: &prompt_tools,
-        skills: &[],
+        workflows: &[],
         dispatcher_instructions: "",
         learned: LearnedContextData::default(),
         visible_tool_names: &NO_FILTER,
@@ -135,7 +135,7 @@ fn datetime_section_includes_timestamp_and_timezone() {
         model_name: "test-model",
         agent_id: "",
         tools: &prompt_tools,
-        skills: &[],
+        workflows: &[],
         dispatcher_instructions: "instr",
         learned: LearnedContextData::default(),
         visible_tool_names: &NO_FILTER,
@@ -180,7 +180,7 @@ fn ctx_with_identity(identity: Option<UserIdentity>) -> PromptContext<'static> {
         model_name: "test-model",
         agent_id: "",
         tools: EMPTY_TOOLS,
-        skills: &[],
+        workflows: &[],
         dispatcher_instructions: "",
         learned: LearnedContextData::default(),
         visible_tool_names: visible,
@@ -322,7 +322,7 @@ fn tools_section_pformat_renders_signature_not_schema() {
         model_name: "test-model",
         agent_id: "",
         tools: &prompt_tools,
-        skills: &[],
+        workflows: &[],
         dispatcher_instructions: "",
         learned: LearnedContextData::default(),
         visible_tool_names: &NO_FILTER,
@@ -365,7 +365,7 @@ fn tools_section_uses_pformat_signature_for_text_dispatchers() {
             model_name: "test-model",
             agent_id: "",
             tools: &prompt_tools,
-            skills: &[],
+            workflows: &[],
             dispatcher_instructions: "",
             learned: LearnedContextData::default(),
             visible_tool_names: &NO_FILTER,
@@ -415,7 +415,7 @@ fn user_memory_section_renders_namespaces_with_headings() {
         model_name: "test-model",
         agent_id: "",
         tools: &prompt_tools,
-        skills: &[],
+        workflows: &[],
         dispatcher_instructions: "",
         learned,
         visible_tool_names: &NO_FILTER,
@@ -494,7 +494,7 @@ fn user_memory_section_returns_empty_when_no_summaries() {
         model_name: "test-model",
         agent_id: "",
         tools: &prompt_tools,
-        skills: &[],
+        workflows: &[],
         dispatcher_instructions: "",
         learned,
         visible_tool_names: &NO_FILTER,
@@ -1142,7 +1142,7 @@ fn for_subagent_builder_injects_user_files_even_when_identity_omitted() {
         model_name: "test-model",
         agent_id: "",
         tools: &prompt_tools,
-        skills: &[],
+        workflows: &[],
         dispatcher_instructions: "",
         learned: LearnedContextData::default(),
         visible_tool_names: &NO_FILTER,
@@ -1188,7 +1188,7 @@ fn for_subagent_builder_injects_user_files_even_when_identity_omitted() {
         model_name: "test-model",
         agent_id: "",
         tools: &prompt_tools,
-        skills: &[],
+        workflows: &[],
         dispatcher_instructions: "",
         learned: LearnedContextData::default(),
         visible_tool_names: &NO_FILTER,
@@ -1265,7 +1265,7 @@ fn prompt_tool_constructors_and_user_memory_skip_empty_bodies() {
         model_name: "model",
         agent_id: "",
         tools: &[],
-        skills: &[],
+        workflows: &[],
         dispatcher_instructions: "",
         learned: LearnedContextData {
             tree_root_summaries: vec![ns_summary("user", "kept"), ns_summary("empty", "   ")],
@@ -1296,7 +1296,7 @@ fn ctx_with_learned(learned: LearnedContextData) -> PromptContext<'static> {
         model_name: "test-model",
         agent_id: "",
         tools: prompt_tools,
-        skills: &[],
+        workflows: &[],
         dispatcher_instructions: "",
         learned,
         visible_tool_names: &NO_FILTER,
@@ -1435,7 +1435,7 @@ fn tools_section_empty_for_native() {
         model_name: "test-model",
         agent_id: "",
         tools: &prompt_tools,
-        skills: &[],
+        workflows: &[],
         dispatcher_instructions: "",
         learned: LearnedContextData::default(),
         visible_tool_names: &NO_FILTER,
@@ -1468,7 +1468,7 @@ fn tools_section_nonempty_for_pformat() {
         model_name: "test-model",
         agent_id: "",
         tools: &prompt_tools,
-        skills: &[],
+        workflows: &[],
         dispatcher_instructions: "",
         learned: LearnedContextData::default(),
         visible_tool_names: &NO_FILTER,
@@ -1503,7 +1503,7 @@ fn tools_section_native_with_dispatcher_instructions_returns_instructions() {
         model_name: "test-model",
         agent_id: "",
         tools: &prompt_tools,
-        skills: &[],
+        workflows: &[],
         dispatcher_instructions: "## Tool Use Protocol\n\nUse native tool calling.",
         learned: LearnedContextData::default(),
         visible_tool_names: &NO_FILTER,

src/openhuman/agent/prompts/render_helpers.rs

@@ -640,7 +640,7 @@ pub fn default_workspace_file_content(filename: &str) -> &'static str {
 /// manufacture a full context when they only need the static text.
 fn empty_prompt_context_for_static_sections() -> PromptContext<'static> {
     static EMPTY_TOOLS: &[PromptTool<'static>] = &[];
-    static EMPTY_SKILLS: &[crate::openhuman::workflows::Workflow] = &[];
+    static EMPTY_WORKFLOWS: &[crate::openhuman::workflows::Workflow] = &[];
     static EMPTY_INTEGRATIONS: &[ConnectedIntegration] = &[];
     // SAFETY: the &HashSet reference must outlive the returned context;
     // a leaked OnceLock-style allocation gives us a permanent 'static
@@ -652,7 +652,7 @@ fn empty_prompt_context_for_static_sections() -> PromptContext<'static> {
         model_name: "",
         agent_id: "",
         tools: EMPTY_TOOLS,
-        skills: EMPTY_SKILLS,
+        workflows: EMPTY_WORKFLOWS,
         dispatcher_instructions: "",
         learned: LearnedContextData::default(),
         visible_tool_names: visible,

src/openhuman/agent/prompts/types.rs

@@ -339,7 +339,7 @@ pub struct PromptContext<'a> {
     /// Id of the agent this prompt is being built for.
     pub agent_id: &'a str,
     pub tools: &'a [PromptTool<'a>],
-    pub skills: &'a [Workflow],
+    pub workflows: &'a [Workflow],
     pub dispatcher_instructions: &'a str,
     /// Pre-fetched learned context (empty when learning is disabled).
     pub learned: LearnedContextData,

src/openhuman/agent/tools/run_workflow.rs

@@ -99,6 +99,14 @@ mod guard {
         Ok(())
     }
 
+    /// Test-only reader for the process-lifetime spawn counter. Used by the
+    /// regression test that asserts a rejected spawn (e.g. unknown workflow
+    /// id) doesn't consume a backstop slot.
+    #[cfg(test)]
+    pub fn total_spawns() -> u64 {
+        TOTAL_SPAWNS.load(Ordering::SeqCst)
+    }
+
     /// Acquire an await slot + re-entrancy lock for `key` (a workflow-id +
     /// inputs fingerprint, or `await:<run_id>` for re-attach). `Err` if too
     /// many awaits are in flight (nesting/fan-out cap) or the same key is
@@ -555,6 +563,41 @@ mod tests {
         super::guard::acquire_await("cap-test-9".to_string()).expect("a freed slot is reusable");
     }
 
+    #[tokio::test]
+    async fn unknown_workflow_id_does_not_burn_a_spawn_slot() {
+        // Regression: a rejected spawn (unknown workflow id) must NOT consume a
+        // slot against the process-lifetime backstop. `account_spawn` runs only
+        // in the `Ok(started)` arm — after `spawn_workflow_run_background`
+        // succeeds — so an unknown id (which fails synchronously) never accounts
+        // a spawn. Without this ordering, an agent retrying a bad id would
+        // exhaust the 500-spawn budget for legitimate runs. Asserts the counter
+        // DELTA is zero (the counter is global + monotonic, so absolute value is
+        // shared with the backstop test — hence the serial lock + delta check).
+        let _s = guard_serial().lock().unwrap();
+        let before = super::guard::total_spawns();
+        let t = RunWorkflowTool::new();
+        // wait_seconds: 0 → fire-and-forget path (no await slot taken); the
+        // unknown id makes the spawn fail before accounting.
+        let res = t
+            .execute(json!({
+                "workflow_id": "definitely-not-a-real-workflow-zzz",
+                "wait_seconds": 0
+            }))
+            .await
+            .expect("Ok(ToolResult)");
+        assert!(res.is_error, "unknown workflow id must return a tool error");
+        assert!(
+            res.output().contains("unknown") || res.output().contains("workflow"),
+            "error should reference the unknown workflow: {}",
+            res.output()
+        );
+        let after = super::guard::total_spawns();
+        assert_eq!(
+            before, after,
+            "a rejected spawn must not increment the spawn backstop counter"
+        );
+    }
+
     #[test]
     fn account_spawn_trips_the_process_backstop() {
         let _s = guard_serial().lock().unwrap();

src/openhuman/agent/triage/escalation.rs

@@ -280,7 +280,7 @@ async fn dispatch_target_agent(agent_id: &str, prompt: &str) -> anyhow::Result<S
         workspace_dir: agent.workspace_dir().to_path_buf(),
         memory: agent.memory_arc(),
         agent_config: agent.agent_config().clone(),
-        skills: Arc::new(agent.skills().to_vec()),
+        workflows: Arc::new(agent.workflows().to_vec()),
         memory_context: Arc::new(None), // Sub-agent queries memory via tools if needed
         session_id: format!("triage-{}", uuid::Uuid::new_v4()),
         channel: "triage".to_string(),

src/openhuman/agent/triage/evaluator.rs

@@ -735,7 +735,7 @@ fn extract_inline_prompt(def: &AgentDefinition) -> Option<String> {
                 model_name: "",
                 agent_id: &def.id,
                 tools: &empty_tools,
-                skills: &[],
+                workflows: &[],
                 dispatcher_instructions: "",
                 learned: LearnedContextData::default(),
                 visible_tool_names: &empty_visible,

src/openhuman/agent_orchestration/ops_tests.rs

@@ -86,7 +86,7 @@ fn parent_context(provider: Arc<dyn Provider>) -> ParentExecutionContext {
         workspace_dir: std::env::temp_dir(),
         memory: Arc::new(NoopMemory),
         agent_config: AgentConfig::default(),
-        skills: Arc::new(Vec::new()),
+        workflows: Arc::new(Vec::new()),
         memory_context: Arc::new(None),
         session_id: "orchestrator-session".to_string(),
         channel: "test".to_string(),

src/openhuman/agent_orchestration/tools/spawn_parallel_agents_tests.rs

@@ -201,7 +201,7 @@ fn parent_context_with_provider(
         workspace_dir: std::env::temp_dir(),
         memory: Arc::new(NoopMemory),
         agent_config,
-        skills: Arc::new(Vec::new()),
+        workflows: Arc::new(Vec::new()),
         memory_context: Arc::new(None),
         session_id: "session-test".into(),
         channel: "test".into(),

src/openhuman/agent_orchestration/tools/spawn_worker_thread.rs

@@ -385,7 +385,7 @@ mod tests {
             channel: "test".into(),
             all_tools: Arc::new(vec![]),
             all_tool_specs: Arc::new(vec![]),
-            skills: Arc::new(vec![]),
+            workflows: Arc::new(vec![]),
             memory_context: std::sync::Arc::new(None),
             connected_integrations: vec![],
             on_progress: None,

src/openhuman/agent_orchestration/tools/tools_e2e_tests.rs

@@ -193,7 +193,7 @@ fn parent_context(
         workspace_dir: workspace_dir.to_path_buf(),
         memory: Arc::new(NoopMemory),
         agent_config: Default::default(),
-        skills: Arc::new(Vec::new()),
+        workflows: Arc::new(Vec::new()),
         memory_context: Arc::new(None),
         session_id: "tools-e2e-session".into(),
         channel: "test".into(),

src/openhuman/agent_registry/agents/archivist/prompt.rs

@@ -52,7 +52,7 @@ mod tests {
             model_name: "test",
             agent_id: "archivist",
             tools: &[],
-            skills: &[],
+            workflows: &[],
             dispatcher_instructions: "",
             learned: LearnedContextData::default(),
             visible_tool_names: &visible,

src/openhuman/agent_registry/agents/code_executor/prompt.rs

@@ -56,7 +56,7 @@ mod tests {
             model_name: "test",
             agent_id: "code_executor",
             tools: &[],
-            skills: &[],
+            workflows: &[],
             dispatcher_instructions: "",
             learned: LearnedContextData::default(),
             visible_tool_names: &visible,

src/openhuman/agent_registry/agents/critic/prompt.rs

@@ -51,7 +51,7 @@ mod tests {
             model_name: "test",
             agent_id: "critic",
             tools: &[],
-            skills: &[],
+            workflows: &[],
             dispatcher_instructions: "",
             learned: LearnedContextData::default(),
             visible_tool_names: &visible,

src/openhuman/agent_registry/agents/crypto_agent/prompt.rs

@@ -19,7 +19,7 @@ pub fn build(ctx: &PromptContext<'_>) -> Result<String> {
         agent_id = ctx.agent_id,
         model = ctx.model_name,
         tool_count = ctx.tools.len(),
-        skill_count = ctx.skills.len(),
+        workflow_count = ctx.workflows.len(),
         "[agent_prompt][crypto_agent] build_start"
     );
 
@@ -77,7 +77,7 @@ mod tests {
             model_name: "test",
             agent_id: "crypto_agent",
             tools: &[],
-            skills: &[],
+            workflows: &[],
             dispatcher_instructions: "",
             learned: LearnedContextData::default(),
             visible_tool_names: EMPTY_VISIBLE.get_or_init(HashSet::new),

src/openhuman/agent_registry/agents/desktop_control_agent/prompt.rs

@@ -47,7 +47,7 @@ mod tests {
             model_name: "test",
             agent_id: "desktop_control_agent",
             tools: &[],
-            skills: &[],
+            workflows: &[],
             dispatcher_instructions: "",
             learned: LearnedContextData::default(),
             visible_tool_names: &visible,

src/openhuman/agent_registry/agents/help/prompt.rs

@@ -51,7 +51,7 @@ mod tests {
             model_name: "test",
             agent_id: "help",
             tools: &[],
-            skills: &[],
+            workflows: &[],
             dispatcher_instructions: "",
             learned: LearnedContextData::default(),
             visible_tool_names: &visible,

src/openhuman/agent_registry/agents/integrations_agent/prompt.rs

@@ -185,7 +185,7 @@ mod tests {
             model_name: "test",
             agent_id: "integrations_agent",
             tools: &[],
-            skills: &[],
+            workflows: &[],
             dispatcher_instructions: "",
             learned: LearnedContextData::default(),
             visible_tool_names: EMPTY_VISIBLE.get_or_init(HashSet::new),

src/openhuman/agent_registry/agents/loader.rs

@@ -467,7 +467,7 @@ mod tests {
                         model_name: "test",
                         agent_id: &def.id,
                         tools: &empty_tools,
-                        skills: &[],
+                        workflows: &[],
                         dispatcher_instructions: "",
                         learned: LearnedContextData::default(),
                         visible_tool_names: &empty_visible,

src/openhuman/agent_registry/agents/markets_agent/prompt.rs

@@ -21,7 +21,7 @@ pub fn build(ctx: &PromptContext<'_>) -> Result<String> {
         agent_id = ctx.agent_id,
         model = ctx.model_name,
         tool_count = ctx.tools.len(),
-        skill_count = ctx.skills.len(),
+        workflow_count = ctx.workflows.len(),
         "[agent_prompt][markets_agent] build_start"
     );
 
@@ -79,7 +79,7 @@ mod tests {
             model_name: "test",
             agent_id: "markets_agent",
             tools: &[],
-            skills: &[],
+            workflows: &[],
             dispatcher_instructions: "",
             learned: LearnedContextData::default(),
             visible_tool_names: EMPTY_VISIBLE.get_or_init(HashSet::new),

src/openhuman/agent_registry/agents/mcp_setup/prompt.rs

@@ -52,7 +52,7 @@ mod tests {
             model_name: "test",
             agent_id: "mcp_setup",
             tools: &[],
-            skills: &[],
+            workflows: &[],
             dispatcher_instructions: "",
             learned: LearnedContextData::default(),
             visible_tool_names: visible,

src/openhuman/agent_registry/agents/morning_briefing/prompt.rs

@@ -67,7 +67,7 @@ mod tests {
             model_name: "test",
             agent_id: "morning_briefing",
             tools: &[],
-            skills: &[],
+            workflows: &[],
             dispatcher_instructions: "",
             learned: LearnedContextData::default(),
             visible_tool_names: visible,

src/openhuman/agent_registry/agents/orchestrator/prompt.rs

@@ -200,7 +200,7 @@ mod tests {
             model_name: "test",
             agent_id: "orchestrator",
             tools: &[],
-            skills: &[],
+            workflows: &[],
             dispatcher_instructions: "",
             learned: LearnedContextData::default(),
             visible_tool_names: EMPTY_VISIBLE.get_or_init(HashSet::new),

src/openhuman/agent_registry/agents/planner/prompt.rs

@@ -58,7 +58,7 @@ mod tests {
             model_name: "test",
             agent_id: "planner",
             tools: &[],
-            skills: &[],
+            workflows: &[],
             dispatcher_instructions: "",
             learned: LearnedContextData::default(),
             visible_tool_names: &visible,

src/openhuman/agent_registry/agents/presentation_agent/prompt.rs

@@ -47,7 +47,7 @@ mod tests {
             model_name: "test",
             agent_id: "presentation_agent",
             tools: &[],
-            skills: &[],
+            workflows: &[],
             dispatcher_instructions: "",
             learned: LearnedContextData::default(),
             visible_tool_names: &visible,

src/openhuman/agent_registry/agents/researcher/prompt.rs

@@ -52,7 +52,7 @@ mod tests {
             model_name: "test",
             agent_id: "researcher",
             tools: &[],
-            skills: &[],
+            workflows: &[],
             dispatcher_instructions: "",
             learned: LearnedContextData::default(),
             visible_tool_names: &visible,

src/openhuman/agent_registry/agents/scheduler_agent/prompt.rs

@@ -47,7 +47,7 @@ mod tests {
             model_name: "test",
             agent_id: "scheduler_agent",
             tools: &[],
-            skills: &[],
+            workflows: &[],
             dispatcher_instructions: "",
             learned: LearnedContextData::default(),
             visible_tool_names: &visible,

src/openhuman/agent_registry/agents/skill_creator/prompt.rs

@@ -101,7 +101,7 @@ mod tests {
             model_name: "test",
             agent_id: "skill_creator",
             tools: &[],
-            skills: &[],
+            workflows: &[],
             dispatcher_instructions: "",
             learned: LearnedContextData::default(),
             visible_tool_names: &visible,

src/openhuman/agent_registry/agents/summarizer/prompt.rs

@@ -52,7 +52,7 @@ mod tests {
             model_name: "test",
             agent_id: "summarizer",
             tools: &[],
-            skills: &[],
+            workflows: &[],
             dispatcher_instructions: "",
             learned: LearnedContextData::default(),
             visible_tool_names: &visible,

src/openhuman/agent_registry/agents/task_manager_agent/prompt.rs

@@ -51,7 +51,7 @@ mod tests {
             model_name: "test",
             agent_id: "task_manager_agent",
             tools: &[],
-            skills: &[],
+            workflows: &[],
             dispatcher_instructions: "",
             learned: LearnedContextData::default(),
             visible_tool_names: &visible,

src/openhuman/agent_registry/agents/tool_maker/prompt.rs

@@ -56,7 +56,7 @@ mod tests {
             model_name: "test",
             agent_id: "tool_maker",
             tools: &[],
-            skills: &[],
+            workflows: &[],
             dispatcher_instructions: "",
             learned: LearnedContextData::default(),
             visible_tool_names: &visible,

src/openhuman/agent_registry/agents/tools_agent/prompt.rs

@@ -53,7 +53,7 @@ mod tests {
             model_name: "test",
             agent_id: "tools_agent",
             tools: &[],
-            skills: &[],
+            workflows: &[],
             dispatcher_instructions: "",
             learned: LearnedContextData::default(),
             visible_tool_names: &visible,

src/openhuman/agent_registry/agents/trigger_reactor/prompt.rs

@@ -52,7 +52,7 @@ mod tests {
             model_name: "test",
             agent_id: "trigger_reactor",
             tools: &[],
-            skills: &[],
+            workflows: &[],
             dispatcher_instructions: "",
             learned: LearnedContextData::default(),
             visible_tool_names: &visible,

src/openhuman/agent_registry/agents/trigger_triage/prompt.rs

@@ -52,7 +52,7 @@ mod tests {
             model_name: "test",
             agent_id: "trigger_triage",
             tools: &[],
-            skills: &[],
+            workflows: &[],
             dispatcher_instructions: "",
             learned: LearnedContextData::default(),
             visible_tool_names: &visible,

src/openhuman/channels/runtime/startup.rs

@@ -322,16 +322,10 @@ pub async fn start_channels(mut config: Config) -> Result<()> {
 
     // Install the triggered-workflow subscriber now that workflows are
     // discovered — otherwise any workflow declaring `triggers:` is silently
-    // ignored in the channel runtime. The handle is parked in a process static
-    // so the RAII SubscriptionHandle isn't dropped (which would cancel it).
-    {
-        use crate::core::event_bus::SubscriptionHandle;
-        use std::sync::OnceLock;
-        static TRIGGERED_WORKFLOW_HANDLE: OnceLock<Option<SubscriptionHandle>> = OnceLock::new();
-        TRIGGERED_WORKFLOW_HANDLE.get_or_init(|| {
-            crate::openhuman::workflows::bus::register_triggered_workflow_subscriber(&skills)
-        });
-    }
+    // ignored. Idempotent + shares a process-global OnceLock with the
+    // `bootstrap_core_runtime` site, so it registers exactly once regardless of
+    // which startup path runs first (web-chat-only cores never reach here).
+    crate::openhuman::workflows::bus::ensure_triggered_workflow_subscriber(&workspace);
 
     // Collect tool descriptions for the prompt
     let mut tool_descs: Vec<(&str, &str)> = vec![

src/openhuman/learning/prompt_sections.rs

@@ -296,7 +296,7 @@ mod tests {
             model_name: "test-model",
             agent_id: "",
             tools: &[],
-            skills: &[],
+            workflows: &[],
             dispatcher_instructions: "",
             learned,
             visible_tool_names,

src/openhuman/memory_tools/prompt.rs

@@ -232,7 +232,7 @@ mod tests {
             model_name: "test",
             agent_id: "test",
             tools: &[],
-            skills: &[],
+            workflows: &[],
             dispatcher_instructions: "",
             learned: LearnedContextData::default(),
             visible_tool_names: &visible,

src/openhuman/workflows/bus.rs

@@ -14,7 +14,7 @@
 use crate::core::event_bus::{subscribe_global, DomainEvent, EventHandler, SubscriptionHandle};
 use crate::openhuman::workflows::Workflow;
 use async_trait::async_trait;
-use std::sync::Arc;
+use std::sync::{Arc, OnceLock};
 
 // ── Trigger pattern ───────────────────────────────────────────────────────────
 
@@ -109,7 +109,7 @@ impl TriggeredWorkflowIndex {
                         let p = TriggerPattern::parse(t);
                         if p.is_none() {
                             log::warn!(
-                                "[skills::triggered] skill '{}': malformed trigger {:?} — skipping",
+                                "[workflows::triggered] skill '{}': malformed trigger {:?} — skipping",
                                 skill.name,
                                 t
                             );
@@ -184,7 +184,7 @@ impl EventHandler for TriggeredSkillSubscriber {
         tracing::debug!(
             domain = event.domain(),
             skills = ?matched,
-            "[skills::triggered] event matches {} skill trigger(s); \
+            "[workflows::triggered] event matches {} skill trigger(s); \
              activation handoff to integration layer pending",
             matched.len()
         );
@@ -213,7 +213,7 @@ pub fn register_triggered_workflow_subscriber(skills: &[Workflow]) -> Option<Sub
         return None;
     }
     log::info!(
-        "[skills::triggered] registering subscriber for {} skill(s) with event triggers (domains: {:?})",
+        "[workflows::triggered] registering subscriber for {} skill(s) with event triggers (domains: {:?})",
         index.len(),
         index.domains()
     );
@@ -222,6 +222,39 @@ pub fn register_triggered_workflow_subscriber(skills: &[Workflow]) -> Option<Sub
     }))
 }
 
+/// Process-global parking spot for the triggered-workflow subscription
+/// handle. The RAII [`SubscriptionHandle`] must outlive the process (dropping
+/// it cancels the subscription), and registration must happen exactly once no
+/// matter how many startup paths reach it.
+static TRIGGERED_WORKFLOW_HANDLE: OnceLock<Option<SubscriptionHandle>> = OnceLock::new();
+
+/// Idempotently install the triggered-workflow subscriber.
+///
+/// Loads workflow metadata from `workspace` and registers the subscriber on the
+/// **first** call; subsequent calls are no-ops (the handle is parked in
+/// [`TRIGGERED_WORKFLOW_HANDLE`] so the RAII guard isn't dropped). Safe to call
+/// from every startup path.
+///
+/// Both [`crate::openhuman::channels::start_channels`] (messaging cores) and
+/// [`crate::core::jsonrpc::bootstrap_core_runtime`] (always-run serve boot)
+/// invoke this. `start_channels` is skipped for web-chat-only desktop installs
+/// (no messaging integration connected) and when
+/// `OPENHUMAN_DISABLE_CHANNEL_LISTENERS=1`; registering from
+/// `bootstrap_core_runtime` too means those cores still honour workflow
+/// `triggers:`. The shared `OnceLock` guarantees a single registration
+/// regardless of which path runs first.
+///
+/// NOTE: the subscriber currently only *matches* triggers and logs — the
+/// activation handoff to the integration layer is still pending (see
+/// [`TriggeredSkillSubscriber::handle`]). Registering on web-chat-only cores
+/// enables matching, not yet activation.
+pub fn ensure_triggered_workflow_subscriber(workspace: &std::path::Path) {
+    TRIGGERED_WORKFLOW_HANDLE.get_or_init(|| {
+        let workflows = crate::openhuman::workflows::load_workflow_metadata(workspace);
+        register_triggered_workflow_subscriber(&workflows)
+    });
+}
+
 /// Legacy no-op retained while call-sites migrate to
 /// [`register_triggered_workflow_subscriber`]. Safe to call multiple times.
 pub fn register_workflow_cleanup_subscriber() {}
@@ -242,6 +275,22 @@ mod tests {
         }
     }
 
+    // ── ensure_triggered_workflow_subscriber (C1 boot-path helper) ───────────
+
+    #[test]
+    fn ensure_triggered_workflow_subscriber_is_idempotent_and_safe() {
+        // Covers the boot-path helper called from both `start_channels` and
+        // `bootstrap_core_runtime`: it loads workflow metadata from the
+        // workspace and registers the subscriber exactly once via the
+        // process-global OnceLock. An empty temp workspace yields no triggered
+        // workflows, so registration resolves to `None`; the call must not
+        // panic and must be safe to repeat (the OnceLock makes the second call
+        // a no-op).
+        let tmp = tempfile::tempdir().expect("tempdir");
+        ensure_triggered_workflow_subscriber(tmp.path());
+        ensure_triggered_workflow_subscriber(tmp.path());
+    }
+
     // ── TriggerPattern::parse ────────────────────────────────────────────────
 
     #[test]

src/openhuman/workflows/inject.rs

@@ -52,7 +52,7 @@
 //!
 //! ## Logging
 //!
-//! Every candidate emits a grep-friendly `[skills:inject]` log line
+//! Every candidate emits a grep-friendly `[workflows:inject]` log line
 //! with `matched=<bool>`, reason, and injected bytes (see
 //! [`render_injection`]). A summary line lives in the caller
 //! (`Agent::turn`).
@@ -265,7 +265,10 @@ fn contains_whole_word(haystack_lower: &str, needle_lower: &str) -> bool {
 
 /// Match installed skills against a user message per the heuristic
 /// documented at the top of this module.
-pub fn match_workflows<'a>(skills: &'a [Workflow], user_message: &str) -> Vec<WorkflowMatch<'a>> {
+pub fn match_workflows<'a>(
+    workflows: &'a [Workflow],
+    user_message: &str,
+) -> Vec<WorkflowMatch<'a>> {
     let mentions = extract_mentions(user_message);
     let mention_set: HashSet<String> = mentions.iter().map(|(n, _)| n.clone()).collect();
     let mention_index = |skill_norm: &str| -> Option<usize> {
@@ -278,7 +281,7 @@ pub fn match_workflows<'a>(skills: &'a [Workflow], user_message: &str) -> Vec<Wo
     let lower_msg = user_message.to_lowercase();
 
     let mut matches: Vec<WorkflowMatch<'a>> = Vec::new();
-    for skill in skills {
+    for skill in workflows {
         let normalised_name = normalise(&skill.name);
         let user_invocable = is_user_invocable(skill);
 
@@ -385,7 +388,7 @@ where
             Some(b) => b,
             None => {
                 log::warn!(
-                    "[skills:inject] matched={} reason={} name={} skipped=body_unavailable",
+                    "[workflows:inject] matched={} reason={} name={} skipped=body_unavailable",
                     false,
                     "body_unavailable",
                     name
@@ -415,7 +418,7 @@ where
         let min_truncated = header_len + footer_trunc_len + 1;
         if remaining < min_truncated {
             log::info!(
-                "[skills:inject] matched={} reason={} name={} skipped=budget_exhausted remaining_bytes={}",
+                "[workflows:inject] matched={} reason={} name={} skipped=budget_exhausted remaining_bytes={}",
                 false,
                 "budget_exhausted",
                 name,
@@ -439,7 +442,7 @@ where
             rendered.push_str(&footer_full);
             let injected = header_len + body.len() + footer_full_len;
             log::debug!(
-                "[skills:inject] matched={} reason={} name={} injected_bytes={} truncated={}",
+                "[workflows:inject] matched={} reason={} name={} injected_bytes={} truncated={}",
                 true,
                 m.reason.as_str(),
                 name,
@@ -469,7 +472,7 @@ where
         truncated_any = true;
         let injected = header_len + truncated_body.len() + footer_trunc_len;
         log::warn!(
-            "[skills:inject] matched={} reason={} name={} injected_bytes={} truncated={} body_bytes_total={} body_bytes_kept={}",
+            "[workflows:inject] matched={} reason={} name={} injected_bytes={} truncated={} body_bytes_total={} body_bytes_kept={}",
             true,
             m.reason.as_str(),
             name,

src/openhuman/workflows/ops_parse.rs

@@ -302,7 +302,7 @@ impl Workflow {
     pub fn read_body(&self) -> Option<String> {
         if self.legacy {
             log::debug!(
-                "[skills:inject] read_body skipped for legacy skill.json skill name={}",
+                "[workflows:inject] read_body skipped for legacy skill.json skill name={}",
                 self.name
             );
             return None;
@@ -312,7 +312,7 @@ impl Workflow {
             Some((_, body, _)) => Some(body),
             None => {
                 log::warn!(
-                    "[skills:inject] read_body failed to parse {} for skill {}",
+                    "[workflows:inject] read_body failed to parse {} for skill {}",
                     path.display(),
                     self.name
                 );

src/openhuman/workflows/preflight.rs

@@ -243,36 +243,36 @@ pub async fn run_github_preflight<P: PreflightProbes>(
     probes: &P,
 ) -> Result<(), GithubGateError> {
     let Some(cfg) = cfg else {
-        tracing::debug!("[skills:preflight] github gate skipped: no [github] block");
+        tracing::debug!("[workflows:preflight] github gate skipped: no [github] block");
         return Ok(());
     };
     if !cfg.required {
-        tracing::debug!("[skills:preflight] github gate skipped: required = false");
+        tracing::debug!("[workflows:preflight] github gate skipped: required = false");
         return Ok(());
     }
 
     // (1) Composio GitHub integration must be connected.
     if !probes.composio_toolkit_active("github").await {
-        tracing::warn!("[skills:preflight] github gate fail: composio_github_missing");
+        tracing::warn!("[workflows:preflight] github gate fail: composio_github_missing");
         return Err(GithubGateError::ComposioGithubMissing);
     }
 
     // (2) git binary present.
     if let Err(e) = probes.git_version().await {
-        tracing::warn!(error = %e, "[skills:preflight] github gate fail: git_binary_missing");
+        tracing::warn!(error = %e, "[workflows:preflight] github gate fail: git_binary_missing");
         return Err(GithubGateError::GitBinaryMissing(e));
     }
 
     // (3a) git user.name set.
     let git_name = probes.git_user_name().await;
     if git_name.is_empty() {
-        tracing::warn!("[skills:preflight] github gate fail: git_user_name_missing");
+        tracing::warn!("[workflows:preflight] github gate fail: git_user_name_missing");
         return Err(GithubGateError::GitUserNameMissing);
     }
     // (3b) git user.email set.
     let git_email = probes.git_user_email().await;
     if git_email.is_empty() {
-        tracing::warn!("[skills:preflight] github gate fail: git_user_email_missing");
+        tracing::warn!("[workflows:preflight] github gate fail: git_user_email_missing");
         return Err(GithubGateError::GitUserEmailMissing);
     }
 
@@ -280,7 +280,7 @@ pub async fn run_github_preflight<P: PreflightProbes>(
     match cfg.identity_match {
         IdentityMatch::None => {
             tracing::debug!(
-                "[skills:preflight] github gate pass (identity_match=none, reachability only)"
+                "[workflows:preflight] github gate pass (identity_match=none, reachability only)"
             );
             Ok(())
         }
@@ -289,12 +289,12 @@ pub async fn run_github_preflight<P: PreflightProbes>(
             // identity — confirms the connection is genuinely usable.
             match probes.composio_identity("github").await {
                 Some(_) => {
-                    tracing::debug!("[skills:preflight] github gate pass (identity_match=any)");
+                    tracing::debug!("[workflows:preflight] github gate pass (identity_match=any)");
                     Ok(())
                 }
                 None => {
                     tracing::warn!(
-                        "[skills:preflight] github gate fail: composio_identity_unresolved (identity_match=any)"
+                        "[workflows:preflight] github gate fail: composio_identity_unresolved (identity_match=any)"
                     );
                     Err(GithubGateError::ComposioIdentityUnresolved)
                 }
@@ -305,7 +305,7 @@ pub async fn run_github_preflight<P: PreflightProbes>(
                 Some(n) => n,
                 None => {
                     tracing::warn!(
-                        "[skills:preflight] github gate fail: composio_identity_unresolved (identity_match=strict)"
+                        "[workflows:preflight] github gate fail: composio_identity_unresolved (identity_match=strict)"
                     );
                     return Err(GithubGateError::ComposioIdentityUnresolved);
                 }
@@ -314,14 +314,14 @@ pub async fn run_github_preflight<P: PreflightProbes>(
                 tracing::debug!(
                     composio = %composio_name,
                     git = %git_name,
-                    "[skills:preflight] github gate pass (identity_match=strict)"
+                    "[workflows:preflight] github gate pass (identity_match=strict)"
                 );
                 Ok(())
             } else {
                 tracing::warn!(
                     composio = %composio_name,
                     git = %git_name,
-                    "[skills:preflight] github gate fail: identity_mismatch"
+                    "[workflows:preflight] github gate fail: identity_mismatch"
                 );
                 Err(GithubGateError::IdentityMismatch {
                     composio_username: composio_name,

src/openhuman/workflows/schemas/handlers.rs

@@ -43,7 +43,9 @@ pub(super) fn handle_workflows_list(params: Map<String, Value>) -> ControllerFut
         );
         let summaries = skills.into_iter().map(WorkflowSummary::from).collect();
         to_json(RpcOutcome::new(
-            WorkflowsListResult { skills: summaries },
+            WorkflowsListResult {
+                workflows: summaries,
+            },
             Vec::new(),
         ))
     })
@@ -231,7 +233,7 @@ pub(super) fn handle_workflows_create(params: Map<String, Value>) -> ControllerF
                 );
                 to_json(RpcOutcome::new(
                     WorkflowsCreateResult {
-                        skill: WorkflowSummary::from(skill),
+                        workflow: WorkflowSummary::from(skill),
                     },
                     Vec::new(),
                 ))
@@ -261,7 +263,7 @@ pub(super) fn handle_workflows_update(params: Map<String, Value>) -> ControllerF
         match create_workflow(workspace.as_path(), create_params) {
             Ok(skill) => to_json(RpcOutcome::new(
                 WorkflowsCreateResult {
-                    skill: WorkflowSummary::from(skill),
+                    workflow: WorkflowSummary::from(skill),
                 },
                 Vec::new(),
             )),
@@ -296,7 +298,7 @@ pub(super) fn handle_workflows_install_from_url(params: Map<String, Value>) -> C
                         url: outcome.url,
                         stdout: outcome.stdout,
                         stderr: outcome.stderr,
-                        new_skills: outcome.new_skills,
+                        new_workflows: outcome.new_skills,
                     },
                     Vec::new(),
                 ))

src/openhuman/workflows/schemas/wire_types.rs

@@ -193,7 +193,7 @@ impl From<Workflow> for WorkflowSummary {
 
 #[derive(Debug, Serialize)]
 pub(super) struct WorkflowsListResult {
-    pub(super) skills: Vec<WorkflowSummary>,
+    pub(super) workflows: Vec<WorkflowSummary>,
 }
 
 #[derive(Debug, Serialize)]
@@ -206,7 +206,7 @@ pub(super) struct WorkflowsReadResourceResult {
 
 #[derive(Debug, Serialize)]
 pub(super) struct WorkflowsCreateResult {
-    pub(super) skill: WorkflowSummary,
+    pub(super) workflow: WorkflowSummary,
 }
 
 #[derive(Debug, Serialize)]
@@ -214,7 +214,7 @@ pub(super) struct WorkflowsInstallFromUrlResult {
     pub(super) url: String,
     pub(super) stdout: String,
     pub(super) stderr: String,
-    pub(super) new_skills: Vec<String>,
+    pub(super) new_workflows: Vec<String>,
 }
 
 #[derive(Debug, Serialize)]

tests/agent_archivist_debug_round21_raw_coverage_e2e.rs

@@ -281,7 +281,7 @@ fn parent_context(workspace: &Path, provider: Arc<ScriptedProvider>) -> ParentEx
         workspace_dir: workspace.to_path_buf(),
         memory: Arc::new(StubMemory),
         agent_config: AgentConfig::default(),
-        skills: Arc::new(Vec::new()),
+        workflows: Arc::new(Vec::new()),
         memory_context: Arc::new(Some("parent memory".to_string())),
         session_id: "round21-parent-session".to_string(),
         channel: "round21-channel".to_string(),

tests/agent_builder_public.rs

@@ -182,7 +182,7 @@ fn builder_applies_defaults_and_exposes_public_accessors() {
     );
     assert_eq!(agent.temperature(), 0.7);
     assert_eq!(agent.workspace_dir(), std::path::Path::new("."));
-    assert!(agent.skills().is_empty());
+    assert!(agent.workflows().is_empty());
     assert!(agent.history().is_empty());
     assert_eq!(agent.agent_config().max_tool_iterations, 10);
 }

tests/agent_harness_leftovers_raw_coverage_e2e.rs

@@ -296,7 +296,7 @@ fn build_agent(
         .model_name("round19-model".to_string())
         .temperature(0.0)
         .workspace_dir(workspace.to_path_buf())
-        .skills(Vec::new())
+        .workflows(Vec::new())
         .auto_save(false)
         .event_context("round19-session", "round19-channel")
         .agent_definition_name("round19_agent")
@@ -320,7 +320,7 @@ fn prompt_context<'a>(
         model_name: "round19-model",
         agent_id: "round19_agent",
         tools,
-        skills: &[],
+        workflows: &[],
         dispatcher_instructions: "dispatcher guidance",
         learned,
         visible_tool_names: visible,
@@ -389,7 +389,7 @@ fn parent_context(workspace: PathBuf, provider: Arc<ScriptedProvider>) -> Parent
         workspace_dir: workspace,
         memory: Arc::new(StubMemory::default()),
         agent_config: agent_config(3),
-        skills: Arc::new(Vec::new()),
+        workflows: Arc::new(Vec::new()),
         memory_context: Arc::new(Some("parent memory context".to_string())),
         session_id: "round19-parent-session".to_string(),
         channel: "round19-channel".to_string(),

tests/agent_harness_public.rs

@@ -138,7 +138,7 @@ fn stub_parent_context() -> ParentExecutionContext {
         workspace_dir: std::path::PathBuf::from("/tmp"),
         memory: Arc::new(StubMemory),
         agent_config: AgentConfig::default(),
-        skills: Arc::new(vec![]),
+        workflows: Arc::new(vec![]),
         memory_context: Arc::new(Some("ctx".into())),
         session_id: "test-session".into(),
         channel: "test-channel".into(),

tests/agent_harness_raw_coverage_e2e.rs

@@ -250,7 +250,7 @@ fn build_agent_with_tools(
         .model_name("coverage-model".to_string())
         .temperature(0.0)
         .workspace_dir(workspace.to_path_buf())
-        .skills(Vec::new())
+        .workflows(Vec::new())
         .auto_save(false)
         .event_context("coverage-session", "coverage-channel")
         .agent_definition_name(agent_name)
@@ -281,7 +281,7 @@ fn parent_context(workspace: PathBuf, provider: Arc<ScriptedProvider>) -> Parent
         workspace_dir: workspace,
         memory: Arc::new(StubMemory),
         agent_config: agent_config(),
-        skills: Arc::new(Vec::new()),
+        workflows: Arc::new(Vec::new()),
         memory_context: Arc::new(Some("parent memory context".to_string())),
         session_id: "parent-session".to_string(),
         channel: "coverage-channel".to_string(),

tests/agent_large_round25_raw_coverage_e2e.rs

@@ -322,7 +322,7 @@ fn parent(workspace_dir: PathBuf, provider: Arc<ScriptedProvider>) -> ParentExec
         workspace_dir,
         memory: Arc::new(StubMemory),
         agent_config: AgentConfig::default(),
-        skills: Arc::new(Vec::new()),
+        workflows: Arc::new(Vec::new()),
         memory_context: Arc::new(Some("round25 inherited parent context".to_string())),
         session_id: "round25-session".to_string(),
         channel: "round25".to_string(),

tests/agent_prompts_subagent_raw_coverage_e2e.rs

@@ -287,7 +287,7 @@ fn parent(workspace: PathBuf, provider: Arc<ScriptedProvider>) -> ParentExecutio
         workspace_dir: workspace,
         memory: Arc::new(StubMemory),
         agent_config: AgentConfig::default(),
-        skills: Arc::new(Vec::new()),
+        workflows: Arc::new(Vec::new()),
         memory_context: Arc::new(Some("parent memory survives when allowed".to_string())),
         session_id: "round18-session".to_string(),
         channel: "round18".to_string(),
@@ -310,7 +310,7 @@ fn prompt_context<'a>(
         model_name: "round18-model",
         agent_id: "round18_agent",
         tools,
-        skills: &[],
+        workflows: &[],
         dispatcher_instructions: "dispatcher rules",
         learned: LearnedContextData {
             reflections: vec!["Prefer direct answers.".to_string(), "   ".to_string()],

tests/agent_round26_raw_coverage_e2e.rs

@@ -240,7 +240,7 @@ fn prompt_context<'a>(
         model_name: "round26-model",
         agent_id: "round26-agent",
         tools,
-        skills: &[] as &[Workflow],
+        workflows: &[] as &[Workflow],
         dispatcher_instructions: "round26 dispatcher instructions",
         learned: LearnedContextData {
             reflections: vec![

tests/agent_session_round24_raw_coverage_e2e.rs

@@ -384,7 +384,7 @@ fn prompt_ctx<'a>(
         model_name: "round24-model",
         agent_id: "round24-agent",
         tools,
-        skills: &[],
+        workflows: &[],
         dispatcher_instructions: "",
         learned,
         visible_tool_names: &NO_FILTER,

tests/agent_session_turn_raw_coverage_e2e.rs

@@ -907,7 +907,7 @@ async fn subagent_runner_parent_context_filters_tools_caps_output_and_reports_er
             max_tool_iterations: 5,
             ..AgentConfig::default()
         },
-        skills: Arc::new(Vec::new()),
+        workflows: Arc::new(Vec::new()),
         memory_context: Arc::new(Some("parent memory context".to_string())),
         session_id: "round17-parent-session".to_string(),
         channel: "round17-parent-channel".to_string(),

tests/calendar_grounding_e2e.rs

@@ -151,7 +151,7 @@ async fn test_integrations_agent_has_current_date_context() -> Result<()> {
         workspace_dir: std::env::temp_dir(),
         memory: Arc::new(StubMemory),
         agent_config: openhuman_core::openhuman::config::AgentConfig::default(),
-        skills: Arc::new(vec![]),
+        workflows: Arc::new(vec![]),
         memory_context: Arc::new(None),
         session_id: "test-session".into(),
         channel: "test".into(),

tests/composio_list_tools_stack_overflow_regression.rs

@@ -347,7 +347,7 @@ async fn drive_subagent() {
         workspace_dir: std::env::temp_dir(),
         memory: Arc::new(StubMemory),
         agent_config: AgentConfig::default(),
-        skills: Arc::new(vec![]),
+        workflows: Arc::new(vec![]),
         memory_context: Arc::new(None),
         session_id: "stack-regression-session".into(),
         channel: "test".into(),

tests/inference_agent_raw_coverage_e2e.rs

@@ -1301,7 +1301,7 @@ fn agent_builder_public_paths_cover_required_fields_defaults_and_filters() {
     );
     assert_eq!(agent.temperature(), 0.7);
     assert_eq!(agent.workspace_dir(), std::path::Path::new("."));
-    assert!(agent.skills().is_empty());
+    assert!(agent.workflows().is_empty());
     assert!(agent.history().is_empty());
     assert_eq!(agent.agent_config().max_tool_iterations, 10);
     assert_eq!(agent.tools_arc().len(), 2);
@@ -3308,7 +3308,7 @@ fn agent_pformat_and_prompt_renderers_cover_public_paths() {
         model_name: "agentic-v1",
         agent_id: "planner",
         tools: &prompt_tools,
-        skills: &skills,
+        workflows: &skills,
         dispatcher_instructions: "Use tool calls when useful.",
         learned,
         visible_tool_names: &visible_tool_names,
@@ -3417,7 +3417,7 @@ fn agent_builtin_prompt_builders_cover_all_registered_archetypes() {
             model_name: "agentic-v1",
             agent_id: builtin.id,
             tools: &prompt_tools,
-            skills: &skills,
+            workflows: &skills,
             dispatcher_instructions: "Use available tools when needed.",
             learned: LearnedContextData::default(),
             visible_tool_names: &visible_tool_names,

tests/json_rpc_e2e.rs

@@ -10480,7 +10480,9 @@ async fn json_rpc_workflows_lifecycle_round_trip() {
     )
     .await;
     let create_result = assert_no_jsonrpc_error(&create, "workflows_create");
-    let wf = create_result.get("skill").expect("skill in create result");
+    let wf = create_result
+        .get("workflow")
+        .expect("workflow in create result");
     // `id` is the on-disk slug (WorkflowSummary maps dir_name → id). The
     // persisted frontmatter `name` is the slug too (create slugifies it).
     assert_eq!(wf.get("id").and_then(Value::as_str), Some("bug-triage"));
@@ -10490,9 +10492,9 @@ async fn json_rpc_workflows_lifecycle_round_trip() {
     let list = post_json_rpc(&rpc_base, 9202, "openhuman.workflows_list", json!({})).await;
     let list_result = assert_no_jsonrpc_error(&list, "workflows_list");
     let workflows = list_result
-        .get("skills")
+        .get("workflows")
         .and_then(Value::as_array)
-        .expect("skills array in list result");
+        .expect("workflows array in list result");
     assert_eq!(workflows.len(), 1, "exactly one workflow after create");
     assert_eq!(
         workflows[0].get("id").and_then(Value::as_str),
@@ -10547,7 +10549,7 @@ async fn json_rpc_workflows_lifecycle_round_trip() {
     let after_result = assert_no_jsonrpc_error(&after, "workflows_list");
     assert!(
         after_result
-            .get("skills")
+            .get("workflows")
             .and_then(Value::as_array)
             .expect("workflows array")
             .is_empty(),

tests/monitor_agent_e2e.rs

@@ -286,7 +286,7 @@ fn build_agent(
         .model_name("monitor-e2e-model".to_string())
         .temperature(0.0)
         .workspace_dir(workspace.to_path_buf())
-        .skills(Vec::new())
+        .workflows(Vec::new())
         .auto_save(false)
         .event_context("monitor-e2e-session", "monitor-e2e-channel")
         .agent_definition_name("orchestrator")

tests/personality_e2e.rs

@@ -70,7 +70,7 @@ fn empty_prompt_context<'a>(workspace_dir: &'a std::path::Path) -> PromptContext
         model_name: "test-model",
         agent_id: "orchestrator",
         tools: &[],
-        skills: &[],
+        workflows: &[],
         dispatcher_instructions: "",
         learned: LearnedContextData::default(),
         visible_tool_names,

tests/tools_agent_credentials_state_raw_coverage_e2e.rs

@@ -369,7 +369,7 @@ fn parent_context(workspace: PathBuf, provider: Arc<ScriptedProvider>) -> Parent
             max_tool_iterations: 3,
             ..Default::default()
         },
-        skills: Arc::new(Vec::new()),
+        workflows: Arc::new(Vec::new()),
         memory_context: Arc::new(Some("parent memory".to_string())),
         session_id: "round16-parent".to_string(),
         channel: "round16-channel".to_string(),
@@ -741,7 +741,7 @@ async fn round16_agent_builder_turn_uses_public_harness_paths() {
         .model_name("round16-model".to_string())
         .temperature(0.0)
         .workspace_dir(harness.workspace.clone())
-        .skills(Vec::new())
+        .workflows(Vec::new())
         .auto_save(false)
         .event_context("round16-session", "round16-channel")
         .agent_definition_name("round16_builder")