Module 07 · Delivery & durability observability
runnableAI-DORA
Measure durability, review tax, and real throughput—not commit spam.
Runnable v0.1 slice available on GitHub — architecture and scope below.
Problem
AI coding agents inflate commit counts and PR volume, hiding code churn, review bottlenecks, and rework. Traditional DORA dashboards track deployment frequency without telling you if half the AI-written lines were reverted two weeks later.
Where it sits
Sits across Git repositories and PR workflows as the delivery and durability observability layer—measuring whether AI-accelerated output actually survives past merge.
AI-DORA reads directly from local Git history and optional GitHub PR metadata without requiring an external database. It distinguishes structural code changes from lockfiles, build outputs, and formatting noise.
Its four Prometheus metrics—30-day code turnover rate, PR review tax ratio, AI agent code share, and Complexity-Adjusted Throughput—expose whether AI tooling is creating durable engineering leverage or review-queue debt.
How it fits
AI-DORA complements Stage Gate, Shadow Versioning, and Smart Router: Stage Gate governs readiness before merge, Smart Router and Evals trace model traffic, and AI-DORA measures long-term code durability across all repositories.
Capabilities
- 30-day code turnover rate tracking lines added 30-60d ago modified or reverted later
- PR review tax ratio comparing review wait duration against total lead time
- AI code share attribution across Co-authored-by trailers and Mirevox trace_id tags
- Complexity-Adjusted Throughput (CAT) filtering lockfiles and generated artifacts
- Prometheus /metrics scrape endpoint, Pushgateway support, and Grafana dashboard
- Drop-in GitHub Action (action.yml) and Docker Compose deployment
Other modules
- Smart RouterModel access gets scattered across provider keys, endpoints, deployment rules, and separate UIs. Even a VS Code assistant and a home chatbot can end up repeating the same integration work.
- Evals & English GovernanceEach model platform has its own eval screens, scoring terms, and policy controls. Teams end up comparing results by hand, while governance rules live in documents that do not map cleanly to provider settings.
- Stage GateAI-assisted projects can jump from an appointment straight into generated code. The missing middle—prerequisites, requirements, decisions, and clear advance criteria—is usually spread across documents and understood by only a few people.
- Shadow VersioningOne project can have its brief in Confluence, approvals in SharePoint, code in Git, and evidence in a testing system. Humans see one piece of work, but the tools expose unrelated histories and technical version language.
- Context EngineAn agent working across 15 repositories can spend more tokens rediscovering the codebase than answering the question. Every tool builds its own partial picture, then throws much of that work away.
- Data LineageOnce a prompt crosses a gateway, it can be hard to explain which provider, region, model, or internal service handled it. Existing lineage tools cover data estates well, but AI request paths often need a smaller, trace-linked view.