agentleFS
sssemilsssemil/SKILLS/skills/brutal/brutal-inf-fix-loop/SKILL.md

brutal-inf-fix-loop

Skill2 starsChanged 4 months agoMITView on GitHub

Review and fix an open pull request without a pass limit until a fresh review has no CRITICAL or MAJOR findings. When material findings exist, drain every severity opportunistically; a final MINOR/NIT-only pass is materially clean and may stop.

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 6 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    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. Distinctive:

    Run an uncapped material-correctness loop.

    line 8
  2. Distinctive:

    When material findings exist, publish and drain all four severities from that pass.

    line 21
  3. Distinctive:

    Stop successfully when a fresh pass has zero CRITICAL and zero MAJOR.

    line 24
  4. Very specific:

    No findings at all: completion_kind: zero_findings.

    line 25
  5. Distinctive:

    Do not impose a pass, elapsed-time, token, cost, commit, or no-progress limit.

    line 29
  6. Distinctive:

    Do not run the whole loop inside one Codex thread.

    line 34
  7. Very specific:

    Run one fresh brutal-pr-review material-convergence pass.

    line 52
  8. Distinctive:

    If CRITICAL+MAJOR is zero, stop with the appropriate completion kind.

    line 53

≈ “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

Brutal Infinite Fix Loop

Run an uncapped material-correctness loop. “Infinite” removes arbitrary pass and no-progress caps; it does not require polishing MINOR/NIT findings forever.

Source Skills

Read ../brutal-pr-review/SKILL.md and ../brutal-pr-finding-fixer/SKILL.md before starting. Follow their safety and provider rules.

Convergence Contract

  • Review every pass for CRITICAL, MAJOR, MINOR, and NIT findings.
  • Continue while the fresh pass contains at least one CRITICAL or MAJOR.
  • When material findings exist, publish and drain all four severities from that pass. Lower-severity findings are worth fixing while a material cycle is already required.
  • Stop successfully when a fresh pass has zero CRITICAL and zero MAJOR.
    • No findings at all: completion_kind: zero_findings.
    • MINOR/NIT only: completion_kind: materially_clean. Put the residuals in the review summary and task record; do not create actionable finding comments for them.
  • Do not impose a pass, elapsed-time, token, cost, commit, or no-progress limit.
  • A hard source-skill guard stops blocked; user interruption stops canceled.

Managed Phase Mode

Do not run the whole loop inside one Codex thread. Obey the supervisor's phase snapshot and perform only that phase:

  • review: run one fresh material-convergence review and return a checkpoint with base/head snapshot, review id, counts by severity, queue counts, and residual MINOR/NIT findings.
  • fix: drain the complete generated queue for that review occurrence, verify, push, and return a checkpoint.

The supervisor derives review -> fix when CRITICAL+MAJOR is nonzero and review -> handoff otherwise. Never choose or request the next phase yourself. Return the exact managed result and exit. Same-thread resume is only for an interrupted attempt in the same phase.

Standalone Or Native Mode

  1. Reuse validated integrations and the exact open pull request when supplied; otherwise resolve them through BRUTAL.md.
  2. Run one fresh brutal-pr-review material-convergence pass.
  3. If CRITICAL+MAJOR is zero, stop with the appropriate completion kind.
  4. Drain every generated severity with brutal-pr-finding-fixer until no unhandled occurrence remains or a hard guard blocks.
  5. Re-resolve the pull request and latest head, even if no commit was made, and return to step 2.

Progress updates contain only pass, base/head snapshot, severity counts, fixed/skipped totals, commits, verification status, and blocker state. Redirect verbose test, diff, and provider output to run-local temporary logs. Report status and duration; on failure include only the last 200 lines or 16 KiB, whichever is smaller.

Final Response

Report integrations, pull request, final base/head snapshot, total passes, completion kind, counts fixed/skipped by severity, residual MINOR/NIT findings, commits, verification, and final review id. On a hard stop, include the exact blocker and remaining generated occurrences; never describe it as clean.

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.