Escape rate measures the proportion of defects that reach production from auto-merged pull requests. It turns automation from a throughput claim into a risk-control decision: increase the Green bucket only when observed failures remain below an explicit tolerance. section 10.5.2 section 10.7.1
Use escape rate to govern progressive automation:
- Track defects reaching production from Green, auto-merged changes. section 10.5.2
- Compare the observed rate with a maximum tolerance before promoting more change patterns to Green. section 10.5.2 section 10.7.1
- Read it alongside auto-approve rate and merge queue wait time; throughput without defect control can hide unacceptable production risk. section 10.7.1
Measured promotion
After six months, auto-merged changes had an escape rate of 0.3%, below a 3% maximum tolerance. A checkout change involving financial logic would still remain Yellow regardless of diff size, preventing a currency-rounding defect from qualifying for Green. section 10.5.2 section 10.7.1
Listing 10.4 Policy-based ownership with triage rules
libs/core: owner: platform-team green: "coverage > 95%, diff < 50 lines, no new dependencies" yellow: "platform-team-oncall reviews" red: "platform-team-lead + security-review" services/checkout: owner: payments-team green: "tests pass, no schema changes, no auth modifications" yellow: "payments-team-member reviews" red: "payments-team-lead reviews" sla: "99.99% - minimum Yellow review for all changes" services/notification: owner: platform-team green: "tests pass, no external API changes, diff < 100 lines" yellow: "any platform-team member reviews" red: "platform-team-lead reviews"
Policy rules make escape-rate control actionable by assigning Green, Yellow, and Red conditions to ownership areas and change types. section 10.5.2
