run-tests

name	run-tests
version	1.0.0
description	Run project tests and report results with failure diagnostics and fix suggestions.
allowed_tools	["shell","file_ops"]
tags	["testing","developer","ci"]

Run Tests

Execute the project's test suite, interpret results, and provide diagnostics for any failures. Support common test frameworks: pytest (Python), Jest (JS/TS), Go test, and cargo test (Rust).

Procedure

1. Detect the test framework by checking for:

   • pyproject.toml or pytest.ini → pytest
   • package.json with "jest" key → Jest
   • go.mod → go test
   • Cargo.toml → cargo test If multiple frameworks are detected, ask the user which to run or run all.

2. Show the exact command before running: shell: <command>

3. Run the test command. Standard commands:

   • pytest: python -m pytest <args> -q
   • Jest: npm test -- --ci
   • Go: go test ./... -v
   • Rust: cargo test

4. Capture both stdout and stderr.

5. Parse the output:

   • Count: passed, failed, skipped, errored
   • Identify each failing test by name
   • Extract the failure message and stack trace

6. For each failing test, provide:

   • Test name: the full test ID
   • Failure type: assertion error, exception, timeout, or other
   • Failure message: the actual vs expected values or exception text
   • Likely cause: a 1-2 sentence diagnosis based on the error
   • Suggested fix: a concrete next step (check a specific line, add a missing mock, update a snapshot, etc.)

7. Report a final summary: overall pass/fail status, counts, and total duration.

Output Format

## Test Run: <project/scope>

Command: `<exact command run>`

### Results

Status: PASSED | FAILED
Passed: N  Failed: N  Skipped: N  Duration: Xs

### Failures

#### Test: test_module::test_function_name
Type: AssertionError
Message: assert 42 == 43
Likely cause: The function returns the cached value from a previous test state.
Fix: Ensure test isolation by resetting the cache in setUp/teardown.

### Summary
<Overall assessment: all green, or N failures requiring attention>

Quality Rules

• Always show the command before running it.
• Do not modify test files or production code to make tests pass without explaining the change and confirming with the user.
• If tests pass, report it clearly: "All N tests passed."
• If a test framework is not detected, report the situation and ask rather than guessing.