Expand description
Common action names used in policy evaluation.
Constants§
- ACCESS_
KNOWLEDGE - Accessing a knowledge base.
- CALL_
LLM - Making an LLM API call.
- COLLABORATE
- Sending inter-agent collaboration messages.
- EXECUTE_
TOOL - Executing a tool (MCP or native).
- SKILL_
GENERATE - Generating a new skill from gap detection (PMPO pipeline).
- SKILL_
MUTATE - Optimizing a skill’s prompt (dspy-rs writes back to SKILL.md).
- SKILL_
PROMOTE - Promoting a generated skill from staging to active.
- SPAWN_
AGENT - Spawning a new agent actor.
- TRACE_
CAPTURE - Capturing execution traces for the learning pipeline.
- VALIDATE_
OUTPUT - Validating an LLM response before it is returned to the user.