Now liveThe Skillselion MCP - thousands of ranked skills, loaded into your agent mid-task. No install.Get it →
paulrberg avatar

Code Polish

  • 9 installs
  • 5 repo stars
  • Updated August 4, 2026
  • paulrberg/dot-agents

Helps with ai & agent building tasks.

About

code-polish is a Claude Code skill for ai & agent building. It helps solo builders move faster with AI-assisted development.

  • code-polish
  • AI & Agent Building
  • AI-coding skill

Code Polish by the numbers

  • 9 all-time installs (skills.sh)
  • +1 installs in the week ending Aug 4, 2026 (Skillselion tracking)
  • Ranked #12,133 of 16,546 AI & Agent Building skills by installs in the Skillselion catalog
  • Data as of Aug 5, 2026 (Skillselion catalog sync)
npx skills add https://github.com/paulrberg/dot-agents --skill code-polish

Add your badge

Show developers this skill is listed on Skillselion. Paste this into your README.

Listed on Skillselion
Installs9
repo stars5
Last updatedAugust 4, 2026
Repositorypaulrberg/dot-agents

What it does

Helps with ai & agent building tasks.

Files

SKILL.mdMarkdownGitHub ↗

Code Polish

Combined simplification and review pipeline. This skill orchestrates two sub-skills in sequence:

1. `code-simplify` — simplify for readability and maintainability 2. `code-review --fix` — review for correctness, security, and quality, auto-applying all fixes

Optimize for one scope resolution, one user-facing report, and no redundant simplify verification.

Scope Resolution

1. Verify repository context: git rev-parse --git-dir. If this fails, stop and tell the user to run from a git repository. 2. If user provides file paths/patterns, a commit/range, or a Resolved scope fenced block with one repo-relative path per line, scope is exactly those targets. 3. Otherwise, scope is only session-modified files. Do not include other uncommitted changes. 4. If there are no session-modified files, fall back to all uncommitted tracked + untracked files:

  • tracked: git diff --name-only --diff-filter=ACMR
  • untracked: git ls-files --others --exclude-standard
  • combine both lists and de-duplicate.

5. Exclude generated/low-signal files unless requested: lockfiles, minified bundles, build outputs, vendored code. 6. If scope still resolves to zero files, report and stop. 7. Normalize the final scope into a Resolved scope fenced block with one repo-relative path per line. Reuse that exact block for both sub-skills instead of asking them to rediscover scope.

Workflow

1) Resolve scope once

  • Apply the "Scope Resolution" rules above.
  • Treat the resulting Resolved scope block as authoritative for all downstream work.
  • Forward user intent, constraints, and risk preferences, but do not forward raw duplicate scope selectors when the resolved block already captures them.

2) Run code-simplify

Invoke the code-simplify skill with:

  • the authoritative Resolved scope block
  • --no-verify
  • --no-report
  • any non-scope user intent that still matters

Tell code-simplify not to broaden or rediscover scope.

3) Run code-review --fix

Invoke the code-review skill with:

  • the same authoritative Resolved scope block
  • --fix
  • any non-scope user intent that still matters

If the user explicitly asks for a speed-first pass over maintainability coverage, you may also append --skip-profile naming. Do not skip the naming profile by default.

4) Final verification

  • Treat code-review's post-fix verification as the final verification summary when it already covers the final touched scope.
  • If verification was skipped, partial, or no longer matches the final diff, run one narrow final verification pass across the final touched scope.
  • Always report skipped checks explicitly.

5) Report

Combine the final state into one summary:

1. Scope: Files and functions touched. 2. Simplifications: Key changes from code-simplify, derived from the actual diff when needed because --no-report was used. 3. Review findings and fixes: Findings and applied fixes from code-review. 4. Verification: Commands run and outcomes. 5. Residual risks: Assumptions or items needing manual review.

Related skills

This week in AI coding

Five minutes, every Monday - the tools, releases and tactics for developers.

unsubscribe anytime.