Expand description
Agent beliefs: a structured store of what the agent “knows” about the user, system, task, and conversation.
Following the “state as a first-class citizen” principle (Hitchhiker’s Guide to Agentic AI, Section 18.6.3), beliefs are part of the agent’s explicit state rather than an implicit property of the conversation history.
Each belief carries a confidence score, source attribution, and reinforcement count, enabling the agent to distinguish speculative inferences from confirmed facts.
Structs§
- Agent
Beliefs - All beliefs the agent currently holds, categorized by domain.
- Belief
- A single belief held by the agent, with provenance and confidence.
Enums§
- Belief
Category - High-level category for a belief.
- Belief
Source - Origin category for a belief.