# AGERE AI DEVELOPMENT GOVERNANCE & SOT RULES

Welcome to the **Agere v2.2** repository. All development, refactoring, and UI/UX work in this project must strictly comply with the canonical design specifications in `/doc/`.

## Active Customization Skills
When working on specific domains, ALWAYS reference the relevant workspace skill:
- **Backend Services & API**: See [`.agents/skills/agere-backend/SKILL.md`](file:///Users/mymac/Documents/Projects/Agere%20ID/Agere/.agents/skills/agere-backend/SKILL.md)
  - References: `doc/09_TECHNICAL_ARCHITECTURE_v2.1.md`, `doc/10_CANONICAL_DATABASE_SCHEMA_v2.1.md`, `doc/11_API_CONTRACT_v2.1.md`, `doc/12_STATE_MACHINES_v2.1.md`.
- **Frontend Applications**: See [`.agents/skills/agere-frontend/SKILL.md`](file:///Users/mymac/Documents/Projects/Agere%20ID/Agere/.agents/skills/agere-frontend/SKILL.md)
  - References: `doc/15_FRONTEND_UX_IMPLEMENTATION_SPEC_v2.1.md`, `doc/27_FRONTEND_APPLICATION_MAP_v2.2.md`.
- **UI/UX & Design System**: See [`.agents/skills/agere-uiux/SKILL.md`](file:///Users/mymac/Documents/Projects/Agere%20ID/Agere/.agents/skills/agere-uiux/SKILL.md)
  - References: `doc/Agere Design System v1.5.md`, `doc/33_AGERE_TALENT_BRAND_FINALIZATION_v1.1.md`, `doc/06_CANONICAL_INVARIANTS_v2.1.md`.

## Core Invariants That Must Never Be Broken (I01–I30)
1. **I01**: HR intent cannot be silently altered by AI.
2. **I06**: Missing evidence ≠ Negative evidence (6 discrete evidence states).
3. **I14 & I30**: AI suggests; only authorized humans make consequential hiring decisions with written rationale.
4. **I15**: Decision Readiness ($DR = J \land M \land C \land P \land S \land F$) is an epistemic gating condition, NEVER a quality score or percentage.
5. **I20**: Every consequential transformation must be cryptographically reconstructable.
6. **I24**: Conflict preservation; conflicting evaluations must never be silently averaged out.
7. **I25 & I26**: Accommodations, complaints, or technical glitches must never become negative candidate signals.