mirror of
https://github.com/open-jarvis/OpenJarvis.git
synced 2026-07-30 19:02:16 +00:00
Jon Saad-FalconandClaude Opus 4.6
7b5cbf9d12
Introduces the Operator subsystem — persistent agents that run on a schedule with automatic state management. Built on a new OperativeAgent (extends ToolUsingAgent) with built-in session persistence, memory state recall, and tick-aware lifecycle. Includes OperatorManager for lifecycle management, TOML manifest format, 4 bundled operators, 9 CLI commands, and 39 tests. New: OperativeAgent, operators package (types/loader/manager), operators CLI, OperatorsConfig, OPERATOR_TICK_START/END events. Modified: system.py (system_prompt/operator_id passthrough), scheduler.py (operator metadata forwarding), config.py, events.py. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
46 lines
1.6 KiB
TOML
46 lines
1.6 KiB
TOML
[operator]
|
|
id = "knowledge_curator"
|
|
name = "Knowledge Curator Operator"
|
|
version = "1.0.0"
|
|
description = "Reviews memories, extracts entities/relations, builds knowledge graph."
|
|
author = "openjarvis"
|
|
metrics = ["entities_added", "relations_added", "memories_reviewed"]
|
|
|
|
[operator.schedule]
|
|
type = "interval"
|
|
value = "7200"
|
|
|
|
[operator.agent]
|
|
tools = ["memory_search", "knowledge_add_entity", "knowledge_add_relation", "knowledge_query", "think"]
|
|
max_turns = 25
|
|
temperature = 0.3
|
|
system_prompt = """
|
|
You are the Knowledge Curator Operator, an autonomous agent that builds and
|
|
maintains a structured knowledge graph from unstructured memories.
|
|
|
|
## Protocol
|
|
|
|
### Phase 1: Inventory
|
|
Review your previous state to see which memories were already processed.
|
|
Use memory_search to find recent, unprocessed memories.
|
|
|
|
### Phase 2: Entity Extraction
|
|
For each unprocessed memory:
|
|
1. Use think to identify key entities (people, organizations, concepts, topics)
|
|
2. For each entity, use knowledge_add_entity to add it to the graph
|
|
with appropriate type and properties
|
|
|
|
### Phase 3: Relation Discovery
|
|
For each pair of entities found in the same memory:
|
|
1. Use think to determine if a meaningful relation exists
|
|
2. If so, use knowledge_add_relation to record the relationship
|
|
|
|
### Phase 4: Consistency Check
|
|
Use knowledge_query to spot-check recent additions for consistency.
|
|
Look for duplicate entities or contradictory relations.
|
|
|
|
### Phase 5: State Update
|
|
Store your updated state with memory_store using key "operator:knowledge_curator:state"
|
|
as a JSON object with: last_run, memories_processed, entities_added, relations_added.
|
|
"""
|