AgentSkillsCN

testing

验证 PyPTO 项目的测试覆盖率并运行测试。当您需要运行测试、检查测试覆盖率，或当用户询问有关测试的问题时，可选用此技能。

SKILL.md
--- frontmatter
name: testing
description: Verify testing coverage and run tests for PyPTO project. Use when running tests, checking test coverage, or when the user asks about testing.

PyPTO Testing Skill

Overview

Build and test the PyPTO project to verify code changes haven't broken anything.

How to Use

  1. •Read agent instructions at .claude/agents/testing/AGENT.md
  2. •Invoke Task tool with subagent_type="testing" (specialized agent)
  3. •Agent will build project and run all tests

Environment Setup

If .claude/skills/testing/testing.env exists: Source it before testing.

bash
[ -f .claude/skills/testing/testing.env ] && source .claude/skills/testing/testing.env

If doesn't exist: Skip and suggest creating one (see testing.env.example).

Testing Workflow

bash
# 1. Activate environment (if testing.env exists)
[ -f .claude/skills/testing/testing.env ] && source .claude/skills/testing/testing.env

# 2. Build project
cmake --build build

# 3. Set PYTHONPATH
export PYTHONPATH=$(pwd)/python:$PYTHONPATH

# 4. Run tests
python -m pytest tests/ut/ -v

Test Commands

bash
# Run all tests
python -m pytest tests/ut/ -v

# Run specific test file
python -m pytest tests/ut/test_ir_basic.py -v

# Run specific test
python -m pytest tests/ut/test_ir_basic.py::test_tensor_expr_creation -v

# Run with coverage
python -m pytest tests/ut/ --cov=pypto_core --cov-report=html

Test Structure

code
tests/ut/
├── core/          # Core functionality
├── ir/            # IR (nodes, expressions, operators, parser)
└── pass/          # Pass manager

Testing Checklist

  • • Project builds without errors
  • • No new compiler warnings
  • • All existing tests pass
  • • New features have tests
  • • Bug fixes have regression tests
  • • Tests in tests/ut/ (not elsewhere)

Common Issues

IssueSolution
ImportError: No module named 'pypto_core'export PYTHONPATH=$(pwd)/python:$PYTHONPATH
Tests fail after code changescmake --build build then re-run
Tests in wrong locationMove to tests/ut/

Output Format

code
## Testing Summary
**Status:** ✅ PASS / ⚠️ WARNINGS / ❌ FAIL

### Build Results
[Compiler output, warnings/errors]

### Test Results
- Total: X | Passed: X | Failed: X | Skipped: X

### Failures
[Failed test details if any]

### Recommendations
[Actions to fix issues]

Decision Criteria

StatusCriteria
PASSAll tests pass, build succeeds, no new warnings
WARNINGSTests pass but new warnings or skipped tests
FAILBuild fails or tests fail

Important Notes

  • •Always rebuild before running tests
  • •Check both build and test output
  • •Look for new warnings even if tests pass
  • •Verify new features have corresponding tests

Related Skills

  • •code-review - Code review (runs in parallel with testing)
  • •git-commit - Complete commit workflow