Reflect agent tool definitions (C6) — the provider-neutral LlmToolkit.Tool
set the agentic reflect loop exposes to the LLM. Faithful port of Hindsight's
engine/reflect/tools_schema.py (get_reflect_tools).
Hierarchy the agent is steered through (curated → consolidated → raw):
search_mental_models → search_observations → recall, with expand to
drill a memory up to its chunk/document, and done to answer.
Definitions are pure data; execution is resolved at loop time by the reflect tool resolver (tenant-scoped) — see the C6 loop module.
Summary
Functions
Build the enabled reflect tools. enabled selects which retrieval tiers are
active (mental models / observations are skipped when a bank has none), mirroring
Hindsight's conditional get_reflect_tools. done and expand are always present.
Functions
@spec build(keyword()) :: [LlmToolkit.Tool.t()]
Build the enabled reflect tools. enabled selects which retrieval tiers are
active (mental models / observations are skipped when a bank has none), mirroring
Hindsight's conditional get_reflect_tools. done and expand are always present.