Search, gather, or inventory local AI-agent histories across runtimes and return a read-only, coverage-manifested Retrieval Bundle. Use for requests such as 'find every session where I touched X', 'pull my last month across agents', 'search my Claude Code, Codex, and OpenCode history', or when another workflow needs re-checkable cross-agent evidence. Do not use for one known file or session; use ha-introspect for analysis or conclusions, and ha-introspect-distill to create a skill.
Distill approved repeatable work into a reusable skill, or an explicitly opted-in rule/config change. Use when asked to "distill this", "turn this workflow into a skill", "turn this workflow into a rule", or when handing off repeatable-work findings from an ha-introspect report. Do not use for the work-pattern audit itself (use ha-introspect), raw cross-agent history retrieval (use ha-agent-retrieve), or a one-off task that is not worth reusing.
Produce an evidence-cited, cross-agent self-report with deterministic usage facets and a rigorously bounded blind-spot audit. Use only when explicitly asked for "my usage insights", "audit my work patterns", "where am I wasting effort", "blind spots", or a periodic "self-retro". Do not use for raw records or cross-agent search (use ha-agent-retrieve), a single-session summary, or creating a skill/rule from a finding (use ha-introspect-distill). Heavyweight and explicit-invocation-only; never run it ambiently.