agentleFS
markoblogomarkoblogo/abvx-agent-skills/skills/overengineering-review/SKILL.md

overengineering-review

Skill16 starsChanged 5 months agoMITView on GitHub

Review code or diffs specifically for needless complexity, replaceable dependencies, dead flexibility, and wrappers over stdlib or native platform behavior. Use when the user asks what can be deleted, simplified, inlined, replaced with stdlib/platform features, or whether a change is over-engineered.

1 other project uses a version of this.

Adopted by

Most recently taken up by

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. Very specific:

    Use minimal-diff-builder when the user wants the simplification findings turned into the smallest correct patch.

    line 16
  2. Very specific:

    Use complexity-optimizer instead when the real question is performance or algorithmic cost rather than architectural bloat.

    line 17
  3. Very specific:

    Use architecture-deepening-review instead when the code is not overbuilt but the boundary design is still wrong.

    line 18
  4. Distinctive:

    no vague "this feels complex" commentary.

    line 65
  5. Distinctive:

    If nothing meaningful should be cut:

    line 77
  6. Distinctive:

    Do not recommend deleting trust-boundary validation, security checks, accessibility basics, or data-loss prevention.

    line 83
  7. Distinctive:

    Do not turn the review into a broad bug hunt.

    line 84
  8. Distinctive:

    Do not apply changes unless the user explicitly asks for implementation after review.

    line 85
  9. Distinctive:

    Treat one focused regression test or self-check as acceptable minimum discipline, not bloat.

    line 86
  10. Distinctive:

    When a normal prose review is needed, convert the ranked list into:

    line 90

≈ “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

Overengineering Review

Review only for unnecessary complexity. This skill does not hunt general correctness issues unless they directly affect the simplification recommendation.

Pair With

  • Use minimal-diff-builder when the user wants the simplification findings turned into the smallest correct patch.
  • Use complexity-optimizer instead when the real question is performance or algorithmic cost rather than architectural bloat.
  • Use architecture-deepening-review instead when the code is not overbuilt but the boundary design is still wrong.

Review Surface

Look for:

  • needless abstractions;
  • replaceable dependencies;
  • dead flexibility;
  • wrappers around stdlib or native platform features;
  • local helpers or files that only forward to something simpler;
  • speculative extension points with one caller or one implementation.

Findings Taxonomy

Use these tags:

  • delete: code or layer can disappear entirely.
  • stdlib: custom logic should be replaced by a standard-library primitive.
  • native: use built-in platform behavior instead of app code or dependency code.
  • existing-dep: repo already ships a dependency that makes the extra code unnecessary.
  • yagni: speculative abstraction or configurability with no present payoff.
  • shrink: same behavior, fewer moving parts.

Workflow

  1. Determine scope:
    • diff review;
    • file review;
    • broad repo audit.
  2. Ignore style nitpicks that do not reduce real complexity.
  3. Rank findings by payoff:
    • dependency removal;
    • file/layer deletion;
    • abstraction collapse;
    • line-count shrink.
  4. For each finding, capture:
    • location;
    • tag;
    • what is overbuilt;
    • what replaces it;
    • why the simpler path still covers the requirement;
    • any safety caveat.
  5. Keep findings review-grade:
    • specific;
    • actionable;
    • minimal prose;
    • no vague "this feels complex" commentary.

Output Format

One finding per line in ranked order:

<file>:L<line> <tag> <what to cut>. <replacement>. <why it still covers the need>.

End with one compact summary line:

net: -<N> lines, -<M> files, -<K> deps possible.

If nothing meaningful should be cut:

Lean already. Ship.

Boundaries

  • Do not recommend deleting trust-boundary validation, security checks, accessibility basics, or data-loss prevention.
  • Do not turn the review into a broad bug hunt.
  • Do not apply changes unless the user explicitly asks for implementation after review.
  • Treat one focused regression test or self-check as acceptable minimum discipline, not bloat.

Final Report

When a normal prose review is needed, convert the ranked list into:

  • findings ordered by impact;
  • top simplification opportunities;
  • residual caveats where the simpler path only works under current scope.

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.