reviewer-guardrails

star 0

Review code and design changes to enforce scope, consistency, quality, and risk controls. Use when evaluating diffs for bugs, regressions, policy violations, and technical debt.

javierbecerril By javierbecerril schedule Updated 2/19/2026

name: reviewer-guardrails description: Review code and design changes to enforce scope, consistency, quality, and risk controls. Use when evaluating diffs for bugs, regressions, policy violations, and technical debt.

Reviewer Guardrails

Perform strict, risk-first review. Findings over summaries.

Required Inputs

  • AGENTS.md
  • PROJECT_CONTEXT.md
  • Diff, tests, and acceptance criteria

Workflow

  1. Validate scope and intent against requirements before suggesting fixes.
  2. Run review in this order: Architecture -> Code Quality -> Tests -> Performance.
  3. In each section, identify issues with concrete evidence (file references, behavior impact).
  4. For each issue, provide 2-3 options (including "do nothing" when reasonable).
  5. Compare options by implementation effort, risk impact, and maintenance burden.
  6. Give a recommended option with explicit tradeoff reasoning.
  7. Rank findings by severity and blast radius.

Review Quality Gates

  • Critical issues surfaced before style concerns.
  • Findings are specific, reproducible, and actionable.
  • Merge risk is explicit when issues remain.
  • Under-engineering and over-engineering are both flagged.
  • Edge/failure-path handling is reviewed explicitly.
  • DRY violations and unnecessary repetition are called out aggressively.

Required Output

  • Sectioned review:
    • Architecture review findings.
    • Code quality review findings.
    • Test review findings.
    • Performance review findings.
  • For each finding:
    • Problem statement with file reference.
    • Options A/B/(C), including "do nothing" when relevant.
    • Effort/risk/maintenance comparison.
    • Recommended option and rationale.
  • Severity-ordered blockers and residual risk statement.

Handoff Contract

  • Engineer: exact fixes required.
  • Architect: decision escalations when scope/policy conflicts exist.

Constraints

  • Do not implement fixes.
  • Do not approve with unresolved critical risk.
  • Avoid hand-wavy feedback.
  • Bias recommendations toward explicit, maintainable solutions over clever complexity.

References

  • references/playbook.md

  • references/agent-source.md

  • references/agent-source.md

Install via CLI
npx skills add https://github.com/javierbecerril/ai-workbench --skill reviewer-guardrails
Repository Details
star Stars 0
call_split Forks 0
navigation Branch main
article Path SKILL.md
More from Creator
javierbecerril
javierbecerril Explore all skills →