kevinslin/agstack/debug-test-failure
Claude Code and agent skill by kevinslin/agstack
Trust tier 3
- community
Source check pending
Checking that the SKILL.md file is still in the repository.
Determine whether test failures come from the current branch.
Install
npx vskill@latest install kevinslin/agstack/debug-test-failureSKILL.md
Debug Test Failure
Overview
Use this skill to answer one question: did the current branch's changes introduce the observed test failure, or did the failure already exist at the baseline before the user's or agent's changes?
Workflow
- Capture the current context.
- Read repository instructions and respect prohibited commands.
- Record the current repo path, branch,
git status --short, and test command. - If the user specified test files, test names, or a test command, use that targeted test command.
- If the user did not specify tests, run the repository's full test suite or the closest documented "all tests" command.
- Identify the comparison baseline.
- Determine "me" from the user's explicit identity if provided; otherwise use
git config user.emailandgit config user.name. - Determine the branch base with the upstream or default branch, for example:
- Inspect branch-only commits in chronological order:
- Select the earliest commit in that branch-only list whose author is not "me".
- If every branch-only commit is authored by "me", use
<base>as the comparison commit. - If the selected non-me commit still includes earlier branch changes, call that out in the final report.
- Create an isolated worktree.
- Follow
dev.worktreesconventions when available: store worktrees under~/.worktrees/<repo>/. - Use a detached worktree so no branch pointer moves:
- Do not place the worktree inside the repository.
- If the path already exists, use a clear suffix such as
debug-test-failure-<short-sha>-2.
- Run comparable tests.
- Run the same selected test command in the baseline worktree.
- If the current worktree has not already run that command, run it there too.
- Keep logs separate, for example
/tmp/debug-test-failure-current.logand/tmp/debug-test-failure-baseline.log. - Install or bootstrap dependencies only as required by repository instructions.
- Do not run unrelated precommit hooks or broad cleanup commands unless the user explicitly asked for them.
- Compare results.
- If the baseline passes and the current worktree fails, report that the current branch likely introduced the failure.
- If both fail with the same failure, report that the failure likely predates the current changes or is environmental/flaky.
- If the baseline fails differently, report that the result is inconclusive and separate the baseline failure from the current failure.
- If either run cannot execute, report the exact blocker and avoid claiming causality.
- Report the outcome.
- Include the current branch, baseline commit, baseline worktree path, test command, and pass/fail result for each worktree.
- State the causality conclusion in one sentence.
- Include the most relevant failure excerpt, not the full log.
- Keep the baseline worktree available until the report is complete; remove it only if the user asked for cleanup or the artifact is clearly unnecessary.
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingactive/debug-test-failure/SKILL.md
Popularity
- Repository stars
- 23
- 7-day trend score
- 0
Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/kevinslin/agstack/debug-test-failure)