Run project tests by detecting the environment and fixing failures.

Process

Identify context:

• If code changes are involved: run jj diff -s first to see changed files; then use jj diff -- path to restrict to specific files/directories

• If the user specified specific files or paths, focus on those

Detect the preferred way to run tests: a. Check existing instructions: README.md , CONTRIBUTING.md , CLAUDE.md , AGENTS.md , GEMINI.md , CODEX.md

b. Check task runners: Makefile , justfile , Taskfile.yml for test target c. Check for scripts in bin/ , .my/bin/ (test, *-test, etc.) d. Check package manager scripts (package.json scripts, etc.) e. Fall back to standard: npm test , pytest , go test ./... , cargo test

Run the detected test command

Analyze and fix any failures

Failure Handling

For any test failures:

• Identify the root cause

• Provide specific fixes

• Explain why the fix works

• Re-run tests to verify

Stop Condition

• If a fix fails twice, stop and ask for guidance.

Output

• Test command used

• Failures and fixes (if any)

• Verification