Expand description
Internal tools – cognitive-only tools for agent self-management.
Internal tools operate inside the cognitive graph, reading from
CognitiveState to help the agent manage its working memory,
constraints, plans, and signals. They are NOT exposed to the outer
execution graph.
Tools receive &CognitiveState (read-only) and return JSON describing
what should be applied. Actual state mutation happens via
CognitiveStateUpdate at the node level,
following Pregel snapshot isolation.
Structs§
- Internal
Tool Registry - Registry of internal tools, keyed by name.
Traits§
- Internal
Tool - A tool available only inside the cognitive graph.