Human & AI Agent as Observers
The state-mutation analysis now recognizes non-deterministic observers — humans and AI agents — alongside processes. The boundary detector, reasoning module, and framework doc all reflect this.
Previously, observer enumeration only prompted for process-based actors (consumers, schedulers, health checkers). But humans react to rendered UI affordances and AI agents react to perceived state values — both can act on stale or intermediate state in ways that deterministic code wouldn't. The existing anti-patterns (AP-4 Boundary State Leak, AP-6 Premature State Transition) already cover the mechanics; the gap was in the checklist prompts, not the pattern catalog.
Changed files:
- CLAUDE.md — boundary detector signal: "another process" → "another actor (process, human, or AI agent)"
- modules/state-mutation.md — observer enumeration adds human users and AI agents with affordance-based reasoning guidance
- docs/reasoning-framework.md — Checklist 1 Step 1 updated to match
Housekeeping
- Fixed stale detector.md references in README and plugin-architecture.md (content lives in CLAUDE.md)
- Fixed /dist-review reference → /dist-check in plugin-architecture.md
- Updated skills directory structure in docs to match actual layout (skills/dist-check/SKILL.md)
- Added MIT LICENSE and .gitignore
Full Changelog: https://github.com/Langerrr/distributed-architect/commits/v0.1.1