agentleFS
muxy-appmuxy-app/muxy/.agents/skills/report-task/SKILL.md

report-task

Skill2.4k starsChanged yesterdayApache-2.0View on GitHub

Turn a bug, feature or chore the user reports into a .plans/tasks file for /tasks-start, after confirming it against the code and reproducing it when needed. Use only when the user runs /report-task.

Trending

More talk lately.about the whole repository

Dec 2025: 0 mentionsJan 2026: 0 mentionsFeb 2026: 0 mentionsMar 2026: 0 mentionsApr 2026: 5 mentionsMay 2026: 4 mentionsJun 2026: 0 mentionsJul 2026: 2 mentionsAug 2026: 0 mentionsSep 2026: 1 mentionOct 2026: 6 mentionsNov 2026: 1 mention
Dec 2025Jun 2026Nov 2026
Mentions a month of the muxy-app/muxy repository that our weekly crawl found. There is no adoption data for this file yet.

Who's talking about muxy-app/muxy

The biggest 5 of 19 mentions linking to the repository that our crawl found. Each line is their own headline.

See all 19

Talked about on GitHub, Hacker News and Bluesky (19 times).

On GitHub · 15

10 more on GitHub

On Bluesky · 1

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 2 are distinctive, and 5 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.

  • 5

    Very specific

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

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

    Write only task files in $T.

    line 22
  2. Very specific:

    Don't switch branches or touch the user's uncommitted changes.

    line 24
  3. Very specific:

    Understand: Read docs/product/README.md, docs/tech/README.md and the task files in $T.

    line 36
  4. Very specific:

    Check it's new: Look for the same thing in $T, in open issues (gh issue list --state open --search '<keywords>') and in recent history on …

    line 40
  5. Distinctive:

    Reproduce when the code doesn't settle it: Use the cheapest proof that works:

    line 49
  6. Very specific:

    Clarify: Ask everything a round would otherwise have to ask: the list in tasks-start Triage step 3, starting at "List everything an agent …

    line 65
  7. Distinctive:

    Never split a task, and never make it depend on another one.

    line 82

≈ “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

Report task

Report: $ARGUMENTS. If it's empty, ask the user what they saw or want and wait for the answer.

Repository root: /Users/saeed/Projects/muxy. Task folder: .plans/tasks (call it $T). Task files use the format in .agents/skills/tasks-start/SKILL.md. Read that section before writing one.

The user is here. The result is a task file that a /tasks-start round can finish alone, or no file because the report isn't a real problem.

Rules

  • Write only task files in $T. Don't change code, commit, push, or touch GitHub issues and PRs.
  • Don't switch branches or touch the user's uncommitted changes. Rounds build from main, so run git fetch and judge the report against origin/main (git show origin/main:<path>, git grep <pattern> origin/main).
  • Reproduce in a detached worktree in the scratchpad (git worktree add --detach <dir> origin/main). Remove it with git worktree remove --force <dir> before you finish, and delete any VMs you created.
  • The report is the user's own, so trust it, but confirm it before writing it up as fact.

Steps

  1. Understand. Read docs/product/README.md, docs/tech/README.md and the task files in $T. Restate the report in one or two sentences: what the user did, what happened and what they expected, or for a feature, what they want to do.
  2. Check it's new. Look for the same thing in $T, in open issues (gh issue list --state open --search '<keywords>') and in recent history on origin/main. If a task covers it, offer to add the new details to that task instead. If it's already fixed, name the commit or PR and stop. Then apply tasks-start's "Task size": a new file is the exception. If it fits an open task, plan to add it there as a new part.
  3. Find it in the code. For a bug, trace the code path and find the cause. For a feature, confirm it's missing and find where it would go. Name files and functions, not line numbers.
  4. Reproduce when the code doesn't settle it. Use the cheapest proof that works:

    • a failing test in the worktree
    • the CLI or server run from the worktree
    • a VM for UI, window or platform behaviour, as tasks-start's "Testing in VMs" describes. Export the worktree, not this checkout: .agents/skills/e2e-test/scripts/e2e prepare <run> <worktree>

    Skip this when the code already proves it, or for a feature. If you can't reproduce it, say what you tried and ask the user for what's missing (version, platform, exact steps, logs) before going on.

  5. Verdict. Tell the user whether it's confirmed, not an issue (works as the docs describe, already fixed, or a setup problem) or unconfirmed, with the evidence. Propose the type, size, its place in the order, and which open task it joins, if any. If it isn't an issue, stop without a file unless the user still wants one.
  6. Clarify. Ask everything a round would otherwise have to ask: the list in tasks-start Triage step 3, starting at "List everything an agent would otherwise have to ask". Include the proposed place in the order, or the task it joins, if the user might disagree. Use AskUserQuestion, up to 4 questions per call, with options and a recommendation. Anything the user defers goes under ## Questions.
  7. Write it. To join an open task, add a ### part under its Problem, Changes and Tests. Otherwise write a new task file:
    • Name: <number>-<slug>.md, numbered as tasks-start describes.
    • Problem: in your own words, with the evidence: version, platform, the steps, what the reproduction showed and the cause in the code.
    • Changes: what to change. For a feature or a user-visible bug, add what the user sees and does when it's done. That's the scenario the round tests against.
    • Tests: turn the reproduction into the test that proves the fix.
    • Decisions: the user's answers as short instructions.
    • Ready: yes once it's confirmed and every question is answered, otherwise no.
    • Never split a task, and never make it depend on another one.
  8. Report. Give the file path, title, type, number, size, ready state and a one-line problem. Say if it joined an existing task. List anything left open.

Order

Most urgent first:

  1. Crash, data loss, security hole, or something that blocks normal use.
  2. A user-visible bug or missing behaviour that hurts, with a workaround.
  3. Minor bugs, polish, small features, and refactors that pay off soon.
  4. Large or optional work.

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.