Metrics for an AI-native SDLC
AI-native SDLC metrics pair a fast leading signal for each play with a lagging quality or reliability outcome. Engineering leaders can instrument the loop from Git, pull requests, CI, deployments, and incident records without buying a new analytics system or replacing the DORA measures they already use.
Read timestamps from git and pull-request metadata. Do not invent a new analytics product first.
| Play | Leading indicator | Lagging indicator | Where it lives |
|---|---|---|---|
Capture intent.md | Time from first conversation to a committed intent.md | Survival rate (accepted into Design vs closed); edits to intent.md after the first spec.md | Git history on the intent home |
Write spec.md | Time from intent.md commit to spec.md commit | spec.md commits after the first plan.md for the same change | Git log |
| Plan mode | Share of changes that merge from the first implementation pass; time from plan approval to merged PR | Rework cycles per change; how often the merged diff still matches plan.md | PR metadata |
CLAUDE.md / rules / AGENTS.md | How often the agent repeats a mistake the file should have caught | Time to first merged PR for a new teammate | Git history of the instruction file; PR history |
| Skills | Time from policy-owner approval to the updated skill merging | PR review findings that cite the policy (should fall toward zero) | PRs on the skill folder |
| Parallel sessions | Concurrent sessions per engineer while review quality holds | Changes merged per engineer per week, read with rework rate | OpenTelemetry / vendor export; PR history |
| Feedback loop | First-pass CI success rate for agent-written changes | Review time per PR; change failure rate | CI; PR metadata; incident tracker |
| Continuous evals | Eval pass rate over time; time for a production incident to become a permanent eval | Regressions caught in CI vs production | Eval suite; incident tracker |
| PR review | Time to first review; share of comments resolved without a human touching the branch | Defects and vulnerabilities caught before merge vs production | GitHub/GitLab; incident tracker |
| Approval-gate hooks | Time spent waiting on each gate | Gate violations reaching production, before vs after hooks | Hook telemetry; incident tracker |
| CI/CD | Share of pipeline failures triaged without paging a human | DORA (deployment frequency, lead time, change fail rate, time to restore) | CI and deploy tooling |
| Close the loop | Time from band breach to an intent.md in triage | Share of findings that become merged fixes; repeat incidents of the same class | Detection script log; PR history |
| Scheduled scans | Share of connected repositories on a schedule; time from finding to patch entering review | Vulnerabilities found by the scan vs production or external report; findings per scan over time | Scan history; incident tracker |