agentleFS
dannys-code-cornerdannys-code-corner/incan/.agents/skills/fix/SKILL.md

fix

Skill73 starsChanged 5 days agoApache-2.0View on GitHub

Fix actionable in-scope findings in the current Incan worktree, usually after /review. Use when the user asks to fix review findings, clean up reported issues, apply in-scope review feedback, or says /fix.

82 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use feature, frontend-code-review and 9 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • bradtraversySkill

    feature

    Manage current feature workflow - start, review, explain or complete

    Also in 10 projects that use this

  • chendongqiSkill

    frontend-code-review

    Trigger when the user requests a review of frontend files (e.g., `.tsx`, `.ts`, `.js`). Support both pending-change reviews and focused file reviews while applying the checklist rules.

    Also in 4 projects that use this

  • GoogleSkill

    pr-creator

    Also in 4 projects that use this

  • PJuniszewskiSkill

    cook

    Feature development with guardrails. Plan → Review → Code → Ship.

    Also in 3 projects that use this

  • vbrevikSkill

    work

    Use when the user says 'plan', 'todo', 'copy plan', 'append plan', 'resume plan', 'priorities', or 'what's next'.

    Also in 4 projects that use this

  • MetaSkill

    roadmap

    Build, continue, inspect, or visualize a source-grounded mathematical roadmap and theorem DAG in an existing Autoform Markdown blueprint. Use for source research, repository-wide inventories of existing and planned mathematics, scope, coverage, milestones, and roadmap articles; do not create repository infrastructure or prove Lean declarations.

    Also in 4 projects that use this

The file

Show its rules and text

We split this skill into the rules it gives an agent

16 rules. 0 are common, 9 are distinctive, and 7 are very specific to this project.

  • 0

    Common

    Most of the 75 edited versions we read keep these.

  • 9

    Distinctive

    This author's own choices. Worth a look.

  • 7

    Very specific

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

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

    Do not use /fix as a substitute for first-pass diagnosis when the problem is still unclear.

    line 22
  2. Very specific:

    Start from the current dirty worktree and the concrete findings already on record from /review, inline review comments, or the user's …

    line 28
  3. Very specific:

    Read .agents/state/review-report.md first when it exists.

    line 29
  4. Distinctive:

    Do not widen scope. Avoid opportunistic refactors beyond what is needed to resolve the findings cleanly.

    line 36
  5. Very specific:

    Update .agents/state/review-report.md as findings move between open, fixed, and blocked.

    line 37
  6. Very specific:

    After edits, run make fmt.

    line 39
  7. Distinctive:

    Run the narrowest relevant verification for the fixes you made.

    line 40
  8. Very specific:

    If make pre-commit fails for reasons unrelated to your fixes, keep the local fixes and report the failure as residual risk in the report …

    line 42
  9. Distinctive:

    If findings remain unresolved, classify each as one of:

    line 43
  10. Distinctive:

    Prefer small direct edits over broad rewrites.

    line 53
  11. Distinctive:

    Keep docs and code aligned.

    line 54
  12. Distinctive:

    Do not add a new split-brain workaround while fixing a split-brain bug.

    line 58
  13. Very specific:

    If make pre-commit fails because of unrelated sandbox or network issues, say so explicitly in .agents/state/review-report.md.

    line 68
  14. Distinctive:

    When consuming an orchestrated report, update:

    line 72
  15. Distinctive:

    Do not invent new findings unless the edit work uncovers a tightly related issue needed to finish the requested fix cleanly.

    line 78
  16. Distinctive:

    If nothing needed fixing, say so directly.

    line 98

“Common” is measured: kept by at least 6 in 10 of the 75 edited versions of this file we could read. “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

Fix — Incan Compiler

Purpose

/fix is the repair half of the review workflow. It consumes concrete findings from /review or obvious unresolved worktree issues and fixes them directly when they are in scope and safe.

The primary source of truth is the persistent report in the current worktree:

  • .agents/state/review-report.md

Use /fix when:

  • /review already identified blockers or warnings,
  • the user says "fix the findings",
  • the user wants review cleanup without a fresh diagnostic pass.

Do not use /fix as a substitute for first-pass diagnosis when the problem is still unclear. Use /review first.

/fix is different from /review: it should consume the current report, not wipe it. If the report is stale or obviously belongs to a previous run, rerun /review instead of trying to repair against bad state.

Workflow

  1. Start from the current dirty worktree and the concrete findings already on record from /review, inline review comments, or the user's instructions.
  2. Read .agents/state/review-report.md first when it exists.
  3. For orchestrated reviews, treat ## Findings as the default task list.
  4. For generic single-review reports, fall back to ## Files only if there is no top-level ## Findings list.
  5. Fix only findings that are:
    • in scope,
    • safe without extra user confirmation,
    • concrete enough to act on.
  6. Do not widen scope. Avoid opportunistic refactors beyond what is needed to resolve the findings cleanly.
  7. Update .agents/state/review-report.md as findings move between open, fixed, and blocked.
  8. Append short timestamped entries to ## Activity as you work. Log each major action instead of relying on a final prose recap.
  9. After edits, run make fmt.
  10. Run the narrowest relevant verification for the fixes you made.
  11. Run make pre-commit.
  12. If make pre-commit fails for reasons unrelated to your fixes, keep the local fixes and report the failure as residual risk in the report file.
  13. If findings remain unresolved, classify each as one of:
  • out of scope,
  • risky without user confirmation,
  • external blocker,
  • separate compiler bug.
  1. Preserve findings even when they are non-blocking or design-oriented. design-tension, maintainability, and docs drift are still findings; they may simply move to a lower-priority fix queue instead of being dropped.

Fixing rules

  • Preserve the user's worktree. Never revert unrelated changes.
  • Prefer small direct edits over broad rewrites.
  • Keep docs and code aligned. If you fix behavior, update user-facing docs that claim the old behavior. If behavior is out of scope, correct the docs instead of inventing implementation.
  • For Rust prose comments, write natural paragraphs and let make fmt wrap them. Do not manually chop prose into short lines.
  • For Markdown, keep prose as natural paragraphs. Do not introduce short-prosed or mechanically chopped text.
  • Do not fix boundary bugs with local-only coverage. If a finding involves import, reexport/facade, package consumer, dependency-owned type, test batch, vocab/desugarer, formatter, generated Rust, Rust metadata, or downstream behavior, add the relevant boundary coverage or document why that boundary is impossible for this fix.
  • Do not add a new split-brain workaround while fixing a split-brain bug. Prefer moving the decision into the canonical planner, metadata surface, registry, or shared helper that all relevant paths already consume.

Verification

/fix should usually run:

  • make fmt
  • narrow relevant tests or checks
  • make pre-commit

If make pre-commit fails because of unrelated sandbox or network issues, say so explicitly in .agents/state/review-report.md. That does not justify leaving local findings unfixed.

The report must remain a structured artifact. Do not replace it with a narrative summary. If ## Activity is missing, repair the report before you finish. For orchestrated runs, the canonical report should stay thin and findings-driven; do not recreate a giant per-file checklist there.

When consuming an orchestrated report, update:

  • ## Findings
  • ## Disagreements when relevant
  • ## Verification

Do not invent new findings unless the edit work uncovers a tightly related issue needed to finish the requested fix cleanly. Classification on findings is for handling, not suppression.

Output format

Produce a concise report with:

## Fix — <subject / branch / files>

### Fixed
<item> — <file>:<line> — <what changed>

### Verification
<command> — <result>

### Residual risk
<anything still unresolved, with classification>

If nothing needed fixing, say so directly.

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.