Product design documents for Moira itself — intents, the audit-record spec, the operating model, the agent/gate model, and the red-team verdict that shaped the architecture. These were authored during discovery (originally in the AI SDLC fixture repo) and moved here so Moira's own documentation lives with Moira.
They are design provenance, not the current API/runtime contract — for the
implemented architecture and decisions, see the ADRs in ../adr/.
Where a note overlaps an ADR, the ADR wins.
| Doc | What it is |
|---|---|
| intent-cockpit.md | INT-MOIRA-cockpit — the product intent (governed AI-SDLC cockpit) + market evidence |
| intent-zdzira.md | INT-MOIRA-zdzira-integration — v0.2+ intent: Zdzira as the ticketing/docs/client-collaboration module |
| func-audit-record.md | FUNC-MOIRA-audit-record — per-step audit record (the defensible core); implemented by AuditRecord in the orchestrator |
| operating-model.md | The 7 pillars of running governed agents safely in production (identity, RBAC, secrets, event log, retries, evals, incident loop) |
| agent-and-gate-model.md | The "Configure" layer: producer/verifier agents + configurable gates (auto/human/hybrid) |
| tech-stack.md | Early tech-stack standards (partly historical — LangGraph plan predates the custom DAG engine of ADR-006) |
| red-team-verdict.md | DEC-MOIRA-001 — adversarial red-team verdict and the pivots it forced |
| v0.1-scope-and-plan.md | Historical — the original v0.1 scope (superseded by the ADRs) |
| references/agent-ecosystem.md | Survey of the coding-agent ecosystem and how Moira maps/borrows from it |
| references/exai-ux-inspiration.md | exAI UX/IA inspiration mapped to Moira |
Reconciliation backlog (overlap with shipped docs, intentionally not merged yet): operating-model ↔ ADR-008; agent-and-gate-model ↔ the engine (
orchestrator/moira_core/models.py); func-audit-record ↔AuditRecord+ ADR-004/005; red-team-verdict ↔../ARCHITECTURE_MUST_FIXES.md.