agentleFS
muxy-appmuxy-app/muxy/.agents/skills/code-and-architecture-improvement/SKILL.md

code-and-architecture-improvement

Skill2.4k starsChanged todayApache-2.0View on GitHub

Make one small refactor that preserves behaviour, only when the user explicitly asks to run this skill. Discussing, reviewing or editing the skill does not run its workflow.

Trending

More talk lately.about the whole repository

Dec 2025: 0 mentionsJan 2026: 0 mentionsFeb 2026: 0 mentionsMar 2026: 0 mentionsApr 2026: 5 mentionsMay 2026: 4 mentionsJun 2026: 0 mentionsJul 2026: 2 mentionsAug 2026: 0 mentionsSep 2026: 1 mentionOct 2026: 6 mentionsNov 2026: 1 mention
Dec 2025Jun 2026Nov 2026
Mentions a month of the muxy-app/muxy repository that our weekly crawl found. There is no adoption data for this file yet.

Who's talking about muxy-app/muxy

The biggest 5 of 19 mentions linking to the repository that our crawl found. Each line is their own headline.

See all 19

Talked about on GitHub, Hacker News and Bluesky (19 times).

On GitHub · 15

10 more on GitHub

On Bluesky · 1

The file

We split this skill into the rules it gives an agent

10 rules. About 0 are common, 7 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈7

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    No bug fixes, features or behaviour changes.

    line 11
  2. Distinctive:

    Stop if the working tree has uncommitted changes.

    line 26
  3. Very specific:

    Read docs/tech/architecture.md, then find at least 3 candidate problems that fit the scope.

    line 28
  4. Distinctive:

    If none is worthwhile, stop without changes.

    line 35
  5. Very specific:

    Create a refactor/ branch and record the starting commit.

    line 36
  6. Distinctive:

    Add any missing tests for the behaviour the refactor touches and confirm they pass on the unchanged code.

    line 37
  7. Very specific:

    Run this platform's scripts/check-*.py in full and fix what it finds.

    line 39
  8. Distinctive:

    Commit, push and open a pull request.

    line 45
  9. Distinctive:

    Wait for CI and any review-bot comments.

    line 47
  10. Distinctive:

    If checks cannot pass or the change outgrows the scope, close the PR if one exists, delete the branch, return to main and report why.

    line 52

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Address one concrete maintenance problem with one refactor, as small as a proper fix allows. Keep new or rewritten logic under about 300 lines. Tests, code that only moves and call sites updated to follow it do not count. Preserve observable behaviour, errors, public interfaces, and persisted/protocol formats. No bug fixes, features or behaviour changes.

Make a proper structural fix, not a patch. No shims, pass-through wrappers, flags, special cases or lint suppressions. Tests may move with the code, but never weaken, skip or delete them to get checks passing. For a flaky test, one that also fails intermittently on the starting commit, rerun it and name it in the PR.

This skill runs unattended and on its own. Run no other skills. Never ask the user or wait for approval. Where you would ask, go ahead if the question is about the refactor. Otherwise choose the option that touches nothing outside this run.

Steps

  1. Stop if the working tree has uncommitted changes. Otherwise switch to main and run git pull.
  2. Read docs/tech/architecture.md, then find at least 3 candidate problems that fit the scope. Prefer structure over polish: misplaced responsibilities, leaky boundaries, duplicated logic. Weigh each by maintenance payoff, risk to behaviour and existing test coverage. Skip rename-only or formatting-only churn, areas touched by open PRs, refactors a closed refactor/ PR already tried, and code whose behaviour this machine's checks cannot prove, such as rendering and input handling. Pick the strongest and explain why it helps. If none is worthwhile, stop without changes.
  3. Create a refactor/ branch and record the starting commit.
  4. Add any missing tests for the behaviour the refactor touches and confirm they pass on the unchanged code. Then make the refactor.
  5. Run this platform's scripts/check-*.py in full and fix what it finds.
  6. Review once. Have a sub-agent read every changed line since the starting commit against AGENTS.md, unchanged behaviour and the scope limits, and report only problems the diff introduces, each with file:line and a concrete failure. Verify each finding, fix the real ones and rerun the checks. Do not review again.
  7. Commit, push and open a pull request. Keep the description short: what the refactor improves and which checks passed.
  8. Wait for CI and any review-bot comments. Apply the comments that are right and keep behaviour unchanged, then push. Squash-merge only after every CI job passes on the final commit and the change still fits the scope limits. Then delete the remote branch and return to main.

If checks cannot pass or the change outgrows the scope, close the PR if one exists, delete the branch, return to main and report why. Do not fix bugs found along the way. List them in the PR or the report.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.