Auto-load domain knowledge at task start. Reads /knowledge/INDEX.md, finds relevant domain folders for the current task, loads rules.md (apply by default) and hypotheses.md (check if testable). Invoke manually when switching domains mid-conversation or when working in a domain you haven't loaded yet.
Load relevant domain knowledge before starting work. This ensures confirmed rules are applied and testable hypotheses are tracked.
knowledge/INDEX.md to see all available domainsrules.md in each relevant domain. These are confirmed patterns (5+ confirmations). Apply them by default without being askedhypotheses.md in each relevant domain. Note which ones could be tested or observed during this taskLoaded: [domain1] (N rules, M hypotheses), [domain2] (N rules, M hypotheses)
Watching: [hypothesis that could be tested today]
/mem-save laterLoad multiple domains when tasks span boundaries. When uncertain, load the domain — better to have context you don't need than miss a rule you should follow.