agentleFS
yeaight7yeaight7/agent-powerups/skills/build-fix-minimal-diff/SKILL.md

build-fix-minimal-diff

Skill6 starsChanged 4 months agoApache-2.0View on GitHub

Use when a build, test, type check, or lint is failing — locally or in CI — and the fix must be the smallest safe diff with no opportunistic refactors or unrelated changes.

  • Installs packages

The file

We split this skill into the rules it gives an agent

12 rules. About 0 are common, 11 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈11

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Run the exact failing command and confirm it fails with the same error.

    line 33
  2. Distinctive:

    If it does not fail locally:

    line 35
  3. Distinctive:

    Check environment differences: tool versions, env vars, installed packages, generated files

    line 36
  4. Very specific:

    Run a clean install (npm ci, pip install -r requirements.txt, etc.) and retry

    line 37
  5. Distinctive:

    Document the environment discrepancy instead of guessing at a fix

    line 38
  6. Distinctive:

    Read the error output from the first line — the first error is usually the root cause.

    line 42
  7. Distinctive:

    Do not start changing code until the root cause is clearly identified

    line 44
  8. Distinctive:

    Do not touch adjacent code, even if it looks wrong — log it as a follow-up

    line 50
  9. Distinctive:

    Do not update test expectations to make tests pass unless the expectations were genuinely wrong (i.e., the code behavior is now correct and …

    line 51
  10. Distinctive:

    Run the exact failing command: it must exit 0

    line 55
  11. Distinctive:

    Run the full test suite to confirm no regressions introduced

    line 56
  12. Distinctive:

    No files modified beyond the narrowest fix

    line 73

≈ “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

Build Fix — Minimal Diff

Fix the failing build, test, or type check. Nothing else.

When to Use

  • A CI build is red and blocking other work
  • Tests pass locally but fail in CI — or vice versa
  • Type errors or lint errors block a commit or pipeline
  • A previous change broke a test that was passing

Core Rule

Touch only what is broken. Fix the root cause. Rerun. Done.

Any unrelated improvement — even a one-line cleanup — is out of scope for this skill.

Inputs

  • The exact failing command (e.g., npm test, cargo build, tsc --noEmit, pytest)
  • The full error output — exact messages and line numbers, not a summary
  • Current branch state (git status --short, git diff --stat)

Workflow

1. Reproduce

Run the exact failing command and confirm it fails with the same error.

If it does not fail locally:

  • Check environment differences: tool versions, env vars, installed packages, generated files
  • Run a clean install (npm ci, pip install -r requirements.txt, etc.) and retry
  • Document the environment discrepancy instead of guessing at a fix

2. Isolate

  • Read the error output from the first line — the first error is usually the root cause; later errors are often cascades
  • Identify the single failure point: file, line number, symbol, or command
  • Do not start changing code until the root cause is clearly identified

3. Fix

  • Apply the minimal change that addresses the identified root cause
  • One change at a time
  • Do not touch adjacent code, even if it looks wrong — log it as a follow-up
  • Do not update test expectations to make tests pass unless the expectations were genuinely wrong (i.e., the code behavior is now correct and the test was testing the old — wrong — behavior)

4. Rerun

  • Run the exact failing command: it must exit 0
  • Run the full test suite to confirm no regressions introduced

5. Report

FIXED: <what was broken — command and error>
CAUSE: <root cause — specific, not vague>
CHANGE: <file:line — what was changed and why>
VERIFIED: <command run> → exit 0
REGRESSIONS: none / [list]
DEFERRED: <adjacent issues noticed but not touched>

Verification

  • Exact failing command exits 0 after the fix
  • Full test suite still passes
  • No files modified beyond the narrowest fix

Failure Modes

  • Fixing a cascade error instead of the root cause — go back to step 2
  • Changing test assertions to match broken implementation — only valid when the test was wrong, never when the code is wrong
  • Multiple changes at once — revert all but the smallest hypothesis and retry
  • "While I'm here" scope creep — log it as a follow-up task, do not include it in this fix

Sources / Inspiration

Inspired by Yeachan-Heo/oh-my-codex prompts/build-fixer.md. Rewritten in Agent Powerups style.

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.