Evaluate ai-agent-rules adoption for HealthReporting

[EduardPetraeus]

Problem

HealthReporting's governance (CLAUDE.md, 560 lines) was built inline as a PoC before ai-agent-rules existed. The original ai-governance-framework repo is now archived. The inherits_from reference has been removed, but no decision exists on whether to adopt the shipped ai-agent-rules v0.1.0 framework.

Proposed Solution

Run a gap analysis comparing ai-agent-rules features against the current inline governance. Document the decision in an ADR (adopt, partial adopt, or skip).

Key questions

• Does ai-agent-rules add value over the existing inline governance?
• What would migration look like (retrofit existing CLAUDE.md sections)?
• Should HealthReporting dogfood the ai-rules check/fix CLI?
• How does ai-agent-rules handle domain-specific rules (health data, Databricks)?
• Is the 200-line CLAUDE.md limit compatible with this project's needs?

References

• DOGFOOD.md — previous dogfood report (score 55%, identified 5 gaps)
• ~/Github repos/ai-agent-rules/ — shipped framework v0.1.0
• docs/AI_GOVERNANCE.md — current 7-layer inline governance

Acceptance Criteria

• Gap analysis: ai-agent-rules features vs current inline governance
• Decision documented in ADR (adopt, partial adopt, or skip)
• If adopting: migration plan with concrete steps