bug — Claude Code skill: what it does, how to install
bug is a Claude Code skill that converts reported defects into actionable bug reports or issue drafts. It checks local logs and code first, asking only for missing environment details.
Apache-2.0 ★ 515 updated 2026-11-17
npx skills add liujh0729-jpg/AI-enterprise-OEM --skill bug -a claude-codeWhat bug does
bug turns reported defects into concise issue drafts that other developers can reproduce or investigate. The agent inspects local logs and workspace code before asking questions, avoiding unnecessary inquiries.
Each report includes the symptom in the title, a minimal reproduction path, expected versus actual results, and confirmed environment parameters. It strictly separates tests run by the agent from user reports. Potential causes and workarounds are cited only with code evidence, omitting speculative severity tags.
Credentials and private session data are filtered out automatically. Generating an issue draft does not authorize publishing it to external trackers without user consent.
What it helps with
- turn defect descriptions into reproducible bug reports
- extract minimal reproduction steps from logs and workspace code
- isolate missing environment parameters such as OS versions or device models
- strip credentials and private session tokens from issue drafts
Example prompts for the agent
- Document the crash occurring when opening the user profile
- Generate a bug report from the latest failure in the logs
- Draft an issue for the network connection timeout defect
How to install bug
1Skills CLI (Vercel)
npx skills add liujh0729-jpg/AI-enterprise-OEM --skill bug -a claude-code2Manually: put the files in place
~/.claude/skills/bug/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/bug/
1Skills CLI (Vercel)
npx skills add liujh0729-jpg/AI-enterprise-OEM --skill bug -a codex2Manually: put the files in place
.agents/skills/bug/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add liujh0729-jpg/AI-enterprise-OEM --skill bug -a cursor2Manually: put the files in place
.cursor/skills/bug/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install liujh0729-jpg/AI-enterprise-OEM mobot-app/.agents/skills/bug2Manually: put the files in place
.github/skills/bug/SKILL.md1Skills CLI (Vercel)
npx skills add liujh0729-jpg/AI-enterprise-OEM --skill bug -a gemini-cli2Platform CLI
gemini skills install https://github.com/liujh0729-jpg/AI-enterprise-OEM --path mobot-app/.agents/skills/bugManually: put the files in place
.devin/skills/bug/SKILL.mdThe legacy .windsurf/skills/ path still works
Claude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about bug
How to install bug in Claude Code?
Run npx skills add liujh0729-jpg/AI-enterprise-OEM --skill bug -a claude-code or save the configuration to ~/.claude/skills/bug/SKILL.md.
What does the bug skill do?
It processes defect reports into structured issue drafts containing minimal reproduction steps and code evidence.
Does bug automatically post issues to external trackers?
No. Drafting a report does not authorize external publishing. The agent requires prior user consent to submit issues.
Source text
SKILL.md · Apache-2.0
--- name: bug description: Turn a reported defect into an actionable bug report or issue draft. Use when the user asks to capture or document a bug; a request to fix broken behavior should proceed to diagnosis and repair. --- Produce a concise report another person can reproduce or investigate. If the user asked for a fix, use the report as working context, continue the repair, and validate the result. 1. Extract the symptom, expected behavior, and reproduction details already supplied. Inspect available logs and relevant code before asking for facts the workspace can provide. 2. Ask only for missing information that materially affects reproduction or diagnosis, such as Android version, host connection type, device model or the triggering input. Continue independent investigation while waiting; label unknowns instead of guessing. 3. Include the symptom in the title, then the smallest known reproduction, expected versus actual result, relevant environment, and impact. Distinguish a reproduction you ran from one reported by the user. 4. Include a cause or workaround only when evidence supports it; label hypotheses and cite the relevant code. Omit empty sections and speculative severity labels. Keep credentials and private session content out of the report. Drafting a report does not itself authorize posting an issue; use any posting authorization already given in the conversation.
© liujh0729-jpg · Apache-2.0 · liujh0729-jpg/AI-enterprise-OEM