Problem
Launchplane needs durable docs before public skills consume agent-facing context so future work preserves the boundary between GitHub planning truth, Launchplane runtime evidence, and optional skill helpers.
Scope
- Document agent authz subjects and safety levels.
- Document redaction, freshness, provenance, and no-plaintext-secret rules.
- Document read models and final aggregation endpoint once stable.
- Document write-intent flows and approval/dry-run expectations.
- Document how
codex-skills should consume Launchplane context without hardcoding private hosts or copying issue bodies.
Acceptance Criteria
- Service boundary, work graph, operator experience, records/secrets, and any new agent-context docs are consistent.
- Docs state GitHub remains planning/code workflow truth.
- Docs state Launchplane compiles product/runtime/evidence context and scoped intents.
- Docs are updated in the same PRs that add or change agent-facing behavior.
Relationships
Problem
Launchplane needs durable docs before public skills consume agent-facing context so future work preserves the boundary between GitHub planning truth, Launchplane runtime evidence, and optional skill helpers.
Scope
codex-skillsshould consume Launchplane context without hardcoding private hosts or copying issue bodies.Acceptance Criteria
Relationships