Simplify & Harden CI

Install

npx skills add pskoett/pskoett-ai-skills/simplify-and-harden-ci

Purpose

• No code mutation in CI
• Review only changed files
• Emit structured findings
• Optionally block merge based on severity gates

simplify-and-harden

Context Limitation (Important)

• Prefer scan/report and merge gating
• Do not auto-apply code changes in CI
• Escalate ambiguous findings to interactive review

Prerequisites

1. GitHub Actions enabled for the repository
2. GitHub CLI authenticated (

   gh auth status

   )

3. gh-aw

   installed locally for authoring/validation:

gh extension install github/gh-aw

1. In GitHub Actions jobs, install the CLI with:

- uses: github/gh-aw/actions/setup-cli@main
  with:
    version: v0.2.0-beta

CI Contract

1. Scope lock: review only files changed in the PR
2. Headless execution: report findings, do not apply patches/refactors
3. Structured output: emit

   simplify_and_harden

   summary payload
4. Gate policy:

   • critical

     : fail check when critical harden findings exist

   • advisory

     (optional): fail check when advisory findings are configured to block

Authoring Workflow (gh-aw)

references/workflow-example.md

.github/workflows

1. Copy

   references/workflow-example.md

   template block into

   .github/workflows/simplify-and-harden-ci.md

   .
2. Compile and validate workflow:

gh aw compile --validate --strict

1. Trigger and push workflow changes:

gh aw run simplify-and-harden-ci --push

1. Check status/logs in GitHub Actions and ensure PR feedback is posted.

Prompt Template (CI)

Run Simplify & Harden in CI (headless mode) for this pull request.

Rules:
1) Review only files changed in this PR.
2) Do not modify repository files.
3) Simplify pass: detect dead code, naming clarity issues, control-flow complexity, unnecessary API surface, and over-abstraction.
4) Harden pass: detect input-validation gaps, injection vectors, auth/authz issues, secret exposure, data leaks, and concurrency risks.
5) Document pass: suggest non-obvious rationale comments as findings (do not edit files).
6) Emit structured YAML under key `simplify_and_harden`, including:
   - simplify findings
   - harden findings (critical/advisory split)
   - summary counts
   - `review_followup_required`
   - learning loop candidates for self-improvement ingestion
7) If blocking policy is enabled and matching findings exist, mark the run as failed.

Recommended Outputs

1. PR comment with concise findings and severity ordering
2. Check run summary with pass/fail reason
3. Machine-readable YAML artifact for downstream automation

Integration with Self-Improvement

simplify_and_harden.learning_loop.candidates

.learnings/LEARNINGS.md

self-improvement