agentleFS
jackchukajackchuka/skills/gh-issue-report/SKILL.md

gh-issue-report

Skill15 starsChanged 9 days agoMITView on GitHub

Investigate and file bug reports on GitHub repositories. Searches existing issues to avoid duplicates, checks contributing guides and issue templates, optionally confirms the bug in source code, and files a well-structured issue. Works on both local and remote repos. Triggers: "file a bug", "report issue", "is there a bug for", "check if reported", "investigate and file", "bug report on", "/gh-issue-report"

The file

Show its rules and text

We split this skill into the rules it gives an agent

16 rules. About 0 are common, 11 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈11

    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:

    If argument includes owner/repo, use it.

    line 32
  2. Distinctive:

    If the current directory is a git repo, offer it as default.

    line 33
  3. Distinctive:

    Bug description: If provided as argument, use it.

    line 36
  4. Distinctive:

    Check if the repo is cloned locally (look for matching remote in git remote -v or check common paths).

    line 40
  5. Distinctive:

    Local: use file system tools (Read, Grep, Glob) for code investigation.

    line 41
  6. Very specific:

    Remote-only: use gh api to fetch repo contents for code investigation.

    line 42
  7. Distinctive:

    Run 2-3 searches with different keyword combinations extracted from the bug description (e.g., key nouns, component names, error messages).

    line 52
  8. Distinctive:

    If a clear duplicate exists: show it to the user and ask whether to proceed or comment on the existing issue.

    line 55
  9. Distinctive:

    If partial matches exist: show them and note the differences.

    line 56
  10. Distinctive:

    If no matches: continue.

    line 57
  11. Very specific:

    If issue templates exist, fetch the relevant one (usually bug_report.md or bug_report.yml):

    line 71
  12. Distinctive:

    If this yields no useful results (no matching files, or too many to be actionable), skip to Phase 5 and draft the issue without code-level …

    line 95
  13. Distinctive:

    Use subagents for parallel investigation when reading multiple files.

    line 106
  14. Very specific:

    Title: follow repo convention (e.g., [bug] ...), keep under 80 chars

    line 121
  15. Distinctive:

    Suggested fix: if root cause is clear enough to propose a direction

    line 128
  16. Very specific:

    Only use labels that exist on the repo.

    line 137

≈ “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

Issue Report

Investigate a bug, confirm it hasn't been reported, optionally verify in source code, and file a structured issue — all in one workflow.

Arguments

  • /gh-issue-report — interactive: asks for repo and bug description
  • /gh-issue-report <owner/repo> <description> — uses arguments directly

Phase 1: Intake

  1. Resolve target repo:
    • If argument includes owner/repo, use it.
    • If the current directory is a git repo, offer it as default.
    • Otherwise, ask.
  2. Bug description: If provided as argument, use it. Otherwise ask:

    What bug are you seeing? Include steps to reproduce if you have them.

  3. Determine repo locality:
    • Check if the repo is cloned locally (look for matching remote in git remote -v or check common paths).
    • Local: use file system tools (Read, Grep, Glob) for code investigation.
    • Remote-only: use gh api to fetch repo contents for code investigation.

Phase 2: Search Existing Issues

  1. Search for duplicates using multiple query strategies in parallel:

    gh issue list -R <repo> --state all --search "<keywords from description>" --limit 20

    Run 2-3 searches with different keyword combinations extracted from the bug description (e.g., key nouns, component names, error messages). Cast a wide net.

  2. Evaluate matches:
    • If a clear duplicate exists: show it to the user and ask whether to proceed or comment on the existing issue.
    • If partial matches exist: show them and note the differences.
    • If no matches: continue.

Phase 3: Check Repo Conventions

  1. Fetch contributing guide and issue templates in parallel:

    # Contributing guide
    gh api 'repos/<repo>/contents/CONTRIBUTING.md' --jq '.content' | base64 -d
    
    # Issue templates
    gh api 'repos/<repo>/contents/.github/ISSUE_TEMPLATE' --jq '.[].name'

    If issue templates exist, fetch the relevant one (usually bug_report.md or bug_report.yml):

    gh api 'repos/<repo>/contents/.github/ISSUE_TEMPLATE/<template>' --jq '.content' | base64 -d
  2. Extract conventions: note title prefix (e.g., [bug]), required labels, required sections, and any special instructions from the contributing guide.

Phase 4: Code Investigation

Adapt investigation depth based on what's achievable. Start lightweight; go deeper only when the lightweight pass yields promising leads.

  1. Lightweight pass — identify relevant files:

    For local repos:

    Use Grep/Glob to find files matching component names, error messages, or keywords from the bug description.

    For remote repos:

    gh api 'repos/<repo>/git/trees/main?recursive=1' --jq '.tree[].path' | grep -iE '<keywords>'

    If this yields no useful results (no matching files, or too many to be actionable), skip to Phase 5 and draft the issue without code-level root cause. Note in the issue body that root cause was not confirmed in code.

  2. Deep investigation — if the lightweight pass found relevant files (< 10 candidates), read them to understand the bug:

    For local repos: use Read tool directly.

    For remote repos:

    gh api 'repos/<repo>/contents/<path>' --jq '.content' | base64 -d

    Use subagents for parallel investigation when reading multiple files. Focus on:

    • Where the relevant state is stored (React state, store, DB, etc.)
    • What triggers the bug (lifecycle, navigation, race condition, etc.)
    • Why the current implementation fails
  3. Assess findings:
    • Root cause confirmed: include it in the issue with file paths and line references.
    • Suspected but unconfirmed: include as hypothesis, clearly labeled.
    • Nothing actionable found: omit code analysis from the issue, describe only the observed behavior.

Phase 5: Draft Issue

  1. Compose the issue following the repo's template and conventions:

    Always include:

    • Title: follow repo convention (e.g., [bug] ...), keep under 80 chars
    • Description: clear, concise explanation of the bug
    • Steps to reproduce: numbered list
    • Expected vs actual behaviour

    Include when available:

    • Root cause: with file paths and line references from Phase 4
    • Suggested fix: if root cause is clear enough to propose a direction
    • Environment: platform, version, relevant config
  2. Infer labels: fetch repo labels and match against the issue content:

    gh label list -R <repo> --json name,description --jq '.[] | "\(.name)\t\(.description)"'

    Only use labels that exist on the repo. If the template specifies a label (e.g., bug), use it.

  3. Present draft to user:

    Here's the issue I'll file. Want me to adjust anything?

    Show the full title, labels, and body. Wait for approval.

Phase 6: File Issue

  1. Create the issue:

    gh issue create -R <repo> \
      --title "<title>" \
      --label "<labels>" \
      --body "$(cat <<'EOF'
    <body>
    EOF
    )"
  2. Report result:
    Filed: <issue-url>

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.