Regression prevention
Skill matt-bat/agent-command-center/skills/regression-prevention
Governance-first agent command center for workflow automation, validation, policy, documentation sync, and execution discipline.
npx -y skills add matt-bat/agent-command-center --skill regression-preventionAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
- 4 stars4 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Implement or review non-trivial code changes while preserving contracts, repository conventions, user work, and affected behavior. Use for logic, refactors, dependencies, APIs, authentication, data, runtime, or tooling changes and for qualitative code review; exclude copy-only edits, explicit scoring requests, and test design that is fully owned by effective-testing-methods.
SKILL.md
5.5 KB, as published. Nobody here has run it
Regression Prevention
Mission
Produce the smallest coherent implementation that satisfies the task without breaking affected behavior.
Operating Modes
implement: inspect, change, validate, and review the final diff.code_review: perform read-only defect analysis with severity-ranked findings.release_readiness: provide change-safety evidence to governance; governance owns the final release decision.
Never turn code_review into weighted scoring unless the user explicitly requests a score.
Core Workflow
1. Preserve the baseline
- Inspect repository status and existing user changes.
- Do not overwrite, revert, reformat, or claim unrelated changes.
- Capture relevant existing behavior and validation state.
- Identify a rollback path proportional to the planned effect.
2. Discover the implementation context
- Read repository instructions and nearby implementation patterns.
- Trace direct callers, data paths, contracts, and user-visible consumers.
- Reuse established naming, types, error semantics, and abstractions.
- Identify the exact behavior that must change and remain stable.
Read implementation-quality.md for detailed construction and final-diff checks.
3. Classify effects
Classify actual planned effects—not topic keywords:
local_low: isolated, reversible local behaviorlocal_significant: multi-module or compatibility-sensitive behaviorexternal_reversible: authorized external state with tested rollbackexternal_irreversible: destructive, migratory, or uncertain recovery
Authentication or migration source code requires domain review. Backup and restore gates activate only when an operation exposes external state or data-loss risk.
4. Implement a coherent slice
- Prefer the simplest implementation that fits existing architecture.
- Keep behavior changes separate from unrelated cleanup.
- Preserve interfaces unless a migration is explicitly authorized.
- Handle failure paths and invalid states deliberately.
- Add comments only for non-obvious rationale.
5. Select validation by changed surface
Use the cheapest layer that can prove the affected behavior, then expand by risk:
- static/type/lint checks for relevant source
- unit tests for deterministic logic
- integration/contract tests for changed boundaries
- browser tests only for changed browser-visible flows
- runtime smoke checks only when an executable path changed
Use Effective Testing Methods when tests must be designed or changed. A missing or blocked layer raises residual risk; it does not justify an unrelated substitute.
6. Review the final diff
- Confirm every changed line serves an authorized requirement.
- Detect accidental formatting churn, dead code, duplicated logic, and stale references.
- Re-check contracts, types, error paths, security boundaries, and performance-sensitive paths where applicable.
- Reconcile every deliverable and prohibition.
7. Report evidence honestly
Report commands run, material results, skipped checks, residual risk, and rollback notes. Never claim no regressions, verified, or release-ready beyond the evidence.
Code Review Mode
For read-only review:
- inspect the diff and relevant surrounding code
- prioritize correctness, security, data loss, compatibility, and missing tests
- report findings by severity with file references and concrete impact
- distinguish proven defects from questions or suggestions
- state when no material findings were found and identify residual test gaps
Read code-review.md for the review checklist.
Compatibility and Migration
For API, schema, event, dependency, or persisted-data changes, read compatibility-and-migrations.md.
Safety Boundaries
- Activating this skill grants no mutation, command, commit, push, migration, deployment, or release authority.
- Respect explicit validation ownership and prohibitions.
- Unknown critical effects fail closed.
- External or irreversible actions require independent safety gates.
Output Contract
For implementation, provide:
- affected behavior and implementation summary
- validation evidence
- residual risk and rollback notes
For code review, provide severity-ranked findings first, then questions and residual coverage gaps.
Quality Gates
- Repository conventions and existing user changes were preserved.
- The implementation is minimal and system-complete.
- Changed contracts and failure paths were checked.
- Validation matches the actual changed surface.
- The final diff contains no unauthorized or unexplained work.
Related Skills
- Diagnose Before Fix: verify an uncertain root cause before implementation.
- Effective Testing Methods: design or amend tests.
- Interdependent Change Planning: coordinate coupled surfaces.
- Skill Governance: own governed mode, gates, and release recommendation.
Gives 0 of the 12 instructions most test skills give
Counted across 964 of the 1,571 authors here whose files we hold, read 2026-08-06
- close the browser when donein 55 of 964, across 12 files
- wait for network idle statein 51 of 964, across 6 files
- launch chromium in headless modein 49 of 964, across 6 files
- use descriptive selectors for elementsin 49 of 964, across 6 files
- run provided scripts with help flag firstin 49 of 964, across 6 files
- add appropriate explicit waitsin 48 of 964, across 5 files
- use bundled scripts as black boxesin 46 of 964, across 3 files
- do not read script source codein 46 of 964, across 3 files
- use sync playwright for scriptsin 46 of 964, across 3 files
- inspect dom before executing actionsin 46 of 964, across 3 files
- run the full test suitein 36 of 964, across 34 files
- write the failing test firstin 25 of 964, across 18 files
Said here and by no other author read
- trace direct callers, data paths, and user-visible consumers
- reuse established naming, types, and abstractions
- classify actual planned effects by risk level
- prefer the simplest implementation fitting existing architecture
- keep behavior changes separate from unrelated cleanup
- handle failure paths and invalid states deliberately
Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once.