// Build and run Skity unit tests with CMake and gtest filtering, including automatic failed-test extraction and machine-readable JSON output. Use when validating specific test suites or running minimal relevant regression tests after code changes.
Guides the capture and optimization of Perfetto traces in benchmarks. Invoke when the user wants to trace performance, analyze bottlenecks, or generate pftrace files.
Check and fix coding style for C++/Obj-C++ and CMake files using the unified code_format_check.py script. Use when you need to verify code formatting before git commit or want to automatically format changed files according to project style guidelines.
| name | test-runner |
| description | Build and run Skity unit tests with CMake and gtest filtering, including automatic failed-test extraction and machine-readable JSON output. Use when validating specific test suites or running minimal relevant regression tests after code changes. |
A skill for building and running unit tests for the Skity project. Supports CMake configuration, building tests, and running specific test cases with gtest filtering. Now includes automatic failure reporting!
# Run all unit tests with automatic failure reporting
python3 tools/test-runner.py
# Run specific test cases with failure reporting
python3 tools/test-runner.py --filter "GeometryTest.*"
# Clean build and run tests with failure reporting
python3 tools/test-runner.py --clean
# Verbose output with failure details
python3 tools/test-runner.py --verbose
# Machine-readable output for agents
python3 tools/test-runner.py --json
$test-runner ... is a conversation trigger syntax for agents, not a shell command.
For terminal usage, run python3 tools/test-runner.py ....
# Run all tests with failure reporting
python3 tools/test-runner.py
# Run only geometry tests
python3 tools/test-runner.py --filter "GeometryTest.*"
# Run tests matching a pattern
python3 tools/test-runner.py --filter "*vector*"
# Clean build and run tests
python3 tools/test-runner.py --clean
# Custom build directory
python3 tools/test-runner.py --build-dir my-build
# Verbose output for debugging
python3 tools/test-runner.py --verbose
# Custom parallel jobs
python3 tools/test-runner.py --parallel 8
# Reuse existing build artifacts (skip configure + build)
python3 tools/test-runner.py --run-only
# Skip only one phase
python3 tools/test-runner.py --no-configure
python3 tools/test-runner.py --no-build
0: tests passed1: test assertion failures2: infrastructure/tooling error (configure/build/runtime failure)3: usage error (for example invalid arguments)--json)The script prints one JSON object with these keys:
mode: "test_runner"status: "ok", "test_failed", or "infra_error"exit_code: normalized script exit codebuild_dir, filter, skip_configure, skip_buildfailed_tests: list of {name, details}passed_test_markers, total_test_markers, test_process_exit_codeduration_msWhen tests fail, the script will:
Example failure output:
❌ Failed test cases detected:
- GeometryTest.VectorAddition
Details: Expected: 42, Actual: 24
- MathTest.MatrixMultiplication
Details: Matrix dimensions don't match
Based on the test structure, you can filter by:
GeometryTest.*MathTest.*GraphicTest.* or specific like PathTest.*ColorTest.*CanvasStateTest.*TextTest.*, TextRunTest.*ArenaAllocatorTest.*, ArrayListTest.*HWBufferLayoutTest.*, HWPipelineKeyTest.*tools/test-runner.pyIf you prefer manual control, you can still use the original commands:
# Configure with CMake
cmake -B build -DSKITY_TEST=ON
# Build the project
cmake --build build --parallel $(nproc 2>/dev/null || sysctl -n hw.ncpu 2>/dev/null || echo 4)
# Run all tests
./build/test/skity_unit_test
# Run specific tests
GTEST_FILTER="GeometryTest.*" ./build/test/skity_unit_test
$(nproc) on Linux or sysctl -n hw.ncpu on macOS for optimal parallel building{build-dir}/test/skity_unit_testGTEST_FILTER environment variable to run specific test cases--gtest_list_tests to see all available tests: ./build/test/skity_unit_test --gtest_list_tests