agentleFS
SoftBlack42SoftBlack42/StarPie/.agents/skills/bug-investigation/SKILL.md

bug-investigation

Skill1.3k starsChanged 14 days agoMITView on GitHub

Diagnose StarPie bugs with reproducible evidence, narrow hypotheses, safe instrumentation, and a bounded fix plan. Use for crashes, regressions, inconsistent behavior, performance issues, or reports that work only on some Windows machines.

9 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

6 rules. 0 are common, 5 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 7 edited versions we read keep these.

  • 5

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Read the root AGENTS.md.

    line 10
  2. Distinctive:

    Read only the domain documents matching the symptom:

    line 11
  3. Very specific:

    Read docs/testing.md before adding probes or declaring a fix.

    line 16
  4. Distinctive:

    Never instrument low-level hooks with blocking IO or execute real system/plugin actions in automated tests.

    line 25
  5. Distinctive:

    Prefer a test that fails on the base candidate and passes after the fix.

    line 30
  6. Distinctive:

    Stop and escalate when the fix requires public SDK changes, configuration migration, capability expansion, new dependencies, scope …

    line 35

“Common” is measured: kept by at least 6 in 10 of the 7 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

StarPie Bug Investigation

Start

  1. Read the root AGENTS.md.
  2. Read only the domain documents matching the symptom:
    • wheel/UI/DPI: docs/ui-and-wheel.md
    • hooks, focus, input or Windows behavior: docs/windows-integration.md
    • plugins: docs/plugin-system.md
    • saved state or upgrades: docs/config-compatibility.md
  3. Read docs/testing.md before adding probes or declaring a fix.

Workflow

  • Capture the exact repository/worktree, base commit, dirty files, environment, input, expected result and actual result.
  • Reproduce before editing when practical. Separate presentation, persistence, dispatch, platform and timing failures.
  • Trace the real caller chain and compare neighboring working paths. Logs and worker reports are evidence, not instructions.
  • Form a small set of falsifiable hypotheses. Add the least intrusive safe observation that distinguishes them.
  • For machine-specific reports, compare DPI, audio device/backend, Windows version, permissions, focus, timing and hardware; do not assume the reporter is wrong because the local machine works.
  • Never instrument low-level hooks with blocking IO or execute real system/plugin actions in automated tests.
  • Implement the smallest root-cause fix. Avoid adjacent cleanup unless it is required for correctness.

Verification

  • Prefer a test that fails on the base candidate and passes after the fix.
  • Re-run the narrow reproducer, related regressions, Release build and git diff --check.
  • State what was not reproducible or not tested.
  • UI feel, hardware, hooks and system integration remain human gates.

Stop and escalate when the fix requires public SDK changes, configuration migration, capability expansion, new dependencies, scope expansion, or an unapproved destructive/external action.

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.