Policy Presets

Policy Presets

Tracking issue: https://github.com/cervantesh/cervo-mutants/issues/11

Preset	Intended use	Operators	Selection	Isolation	Reports
ci-fast	Pull-request smoke and changed-code feedback.	conservative-fast	coverage with prefilter	overlay	summary, JSON, JUnit
ci-balanced	Regular CI where the test suite cost is acceptable.	conservative	coverage with prefilter	overlay	summary, JSON, JUnit
comparison-safe	External tool and multi-repo apples-to-apples calibration.	gremlins-compatible	package	overlay	summary, JSON, JUnit
nightly	Scheduled deeper signal without campaign noise.	default	coverage with prefilter	overlay	summary, JSON, JUnit, HTML
campaign	Manual or scheduled deep mutation campaign.	aggressive	package	temp-workdir	summary, JSON, JUnit, HTML

Comparison Safe

Use comparison-safe for external tool comparisons and multi-repo calibration:

cervomut run . --policy comparison-safe

This preset is intentionally bounded so one difficult repository does not turn the whole study into timeout-only noise:

• mutators.profile: gremlins-compatible
• execution.isolation: overlay
• execution.workers: capped at 2
• execution.budget: 10m
• tests.timeout: 20s
• limits.sample: deterministic
• limits.max_mutants: 250 when no explicit cap is set
• reports: summary, JSON, JUnit
• partial artifacts: partial-summary.json and partial-mutation-report.json

This is the stable comparison lane, not the broadest quality campaign.

Governance Defaults

• ci.fail_under stays at 0; adoption is baseline-first.
• Baseline regression and new survivor policies stay enabled.
• Quarantine requires owner, issue, reason, expiry, and renewal limits.
• Suppression is stricter than quarantine: action: suppress requires evidence: confirmed and at least one reviewer.
• Mutation history is enabled by default at .cervomut/history.json.