agentleFS
whinstudiowhinstudio/skills/cmp-gates/SKILL.md

cmp-gates

Skill0 starsChanged 3 days agoMITView on GitHub

Use to run the finish gates of this Compose Multiplatform project — before saying a change is done, after a screen or module is in place, or before a commit — in a separate subagent that returns a short pass/fail report instead of the Gradle log, fixing only mechanical failures that have a written recipe. Runs in a subagent that does not see this conversation: pass what it needs as the argument.

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 2 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Read before running: ../cmp-verify/SKILL.md sections 1 and 3 (how to read the output.

    line 25
  2. Distinctive:

    Stop at the first red gate and go to section 3.

    line 36
  3. Very specific:

    Never: @Suppress, baselines, detekt config changes, version or wrapper changes, try/catch to hide a crash, deleted or skipped tests.

    line 49
  4. Distinctive:

    No raw log, no command transcript.

    line 66

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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

Gates

You run in your own subagent. The session that called you sees only your final report, not your tool output. You did not see its conversation: everything you need is in the project files.

1. Input

Caller's request: "$ARGUMENTS"

  • The request above says which gates to run, in any wording: a name (android, ios, detekt, tests), a sentence ("only the Android gate") or a gate's command. Run exactly those, nothing else, and leave the others out of the report. "" means all gates.
  • The gate commands are in CLAUDE.md under Definition of done, in order. Run only the gates listed there: a project without detekt has no detekt gate, and that is not a failure.
  • Read before running: ../cmp-verify/SKILL.md sections 1 and 3 (how to read the output; the test gate counts tests and zero is red) and, when a gate is red, ../cmp-verify/references/failures.md. These paths are relative to this skill's folder.

2. Run

  • With an argument, run only the requested gates, even when the list in CLAUDE.md or cmp-verify says to run every gate in order: those rules are for a run without an argument. tests alone means only the test command, not the Android and iOS gates before it.
  • Each gate as written, as its own command, in the foreground, with a long Bash timeout (up to ten minutes). ./gradlew build is never used.
  • Stop at the first red gate and go to section 3; continue with the next requested gate once it is green.

3. Red gate: fix only what has a recipe

  • Match the first real error line against ../cmp-verify/references/failures.md (and for detekt, ../cmp-detekt/SKILL.md sections 2–4: auto-correct, then the listed fixes).
  • A match whose fix is mechanical — an import, the Res package or a resource key import, a Koin or KSP annotation, @Serializable on a destination, detekt formatting — is yours: fix it, following ../cmp-code-rules/SKILL.md, and rerun the same gate. At most 3 rounds per gate.
  • Everything else is reported, not fixed: no recipe, a fix that changes behaviour or logic, a detekt finding that needs a function split or a rename across files, any failing test (never edit a test or the code under test), anything still red after 3 rounds.
  • Never: @Suppress, baselines, detekt config changes, version or wrapper changes, try/catch to hide a crash, deleted or skipped tests.

4. Report (your final message, nothing else)

Gates: Android ✅ · iOS ✅ · detekt ❌ · Tests ✅ (42 tests)
Fixed (mechanical): feature/x/data/src/commonMain/.../XRepository.kt:12 — missing import kotlinx.coroutines.IO
Open:
- detekt LongMethod — feature/x/presentation/.../XContent.kt:88 — needs the function split (logic)
  • One line of gate results: each gate run, ✅ or ❌, the test count for the test gate; gates not listed in CLAUDE.md are left out.
  • Fixed: file:line and one line per fix. Omit when empty.
  • Open: per item the gate, file:line, the first error line shortened to one line, and why you did not fix it. Omit when empty.
  • No raw log, no command transcript.

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.