菜单
Complex pixi task workflows and orchestration. Use when building task dependency chains, configuring caching with inputs/outputs, creating parameterized tasks, or setting up CI pipelines—e.g., "pixi task depends-on", "task caching for build automation", "multi-environment test matrices".
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/diegosouzapw/awesome-omni-skill --skill pixi-tasks命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
基于 SOC 职业分类
正在显示 SKILL.md
| name | pixi-tasks |
| description | Complex pixi task workflows and orchestration. Use when building task dependency chains, configuring caching with inputs/outputs, creating parameterized tasks, or setting up CI pipelines—e.g., "pixi task depends-on", "task caching for build automation", "multi-environment test matrices". |
Complex task patterns: dependency chains, caching with inputs/outputs, cross-environment workflows.
[tasks]
# Simple command
test = "pytest -v"
# With working directory
build = { cmd = "make", cwd = "src" }
# Multiple steps
setup = "mkdir -p build && cmake -B build"
YOU MUST use depends-on for execution order. Tasks without explicit dependencies execute in undefined order and will fail in CI pipelines. No exceptions.
[tasks]
configure = "cmake -B build"
build = { cmd = "cmake --build build", depends-on = ["configure"] }
test = { cmd = "ctest", depends-on = ["build"] }
start = { cmd = "./build/app", depends- = [] }
pixi run start # Runs: configure → build → start
Every time you have multiple upstream dependencies, use this pattern. Tasks with multiple depends-on entries wait for ALL dependencies to complete before executing.
[tasks]
fetch-data = "curl -o data.csv ..."
fetch-model = "curl -o model.pkl ..."
train = { cmd = "python train.py", depends-on = ["fetch-data", "fetch-model"] }
evaluate = { cmd = "python eval.py", depends-on = ["train"] }
Execution: fetch-data and fetch-model complete before train; order is dependency-driven.
Cache results based on inputs/outputs:
[tasks]
# Cache by input file
run = { cmd = "python main.py", inputs = ["main.py"] }
# Cache by inputs and outputs
build = {
cmd = "cargo build --release",
inputs = ["src/**/*.rs", "Cargo.toml"],
outputs = ["target/release/myapp"]
}
# Download caching (by output existence)
download = {
cmd = "curl -o data.csv https://example.com/data.csv",
outputs = ["data.csv"]
}
Cache invalidates immediately when ANY of these conditions are met:
Caches without proper inputs/outputs are always stale. Define both or accept that your tasks will re-run unnecessarily.
Define tasks that behave differently per platform using [target.<platform>.tasks]. Pixi automatically selects the correct implementation based on the current platform.
# Target-agnostic tasks that depend on platform-specific setup
task = { cmd = "echo 'Running main task'", depends-on = ["setup"] }
[target.linux-64.tasks]
setup = { cmd = "echo 'Linux setup'", description = "Setup for Linux" }
[target.osx-arm64.tasks]
setup = { cmd = "echo 'macOS setup'", description = "Setup for macOS" }
pixi run task # Runs the correct 'setup' for your platform, then 'task'
Combine with [target.<platform>.dependencies] to install packages only on specific platforms:
# Install stow only on Linux (via conda-forge)
[target.linux-64.dependencies]
stow = "*"
# Target-agnostic tasks
task = { cmd = "stow --version", depends-on = ["check-stow"] }
# Platform-specific setup tasks (same name, different implementations)
[target.linux-64.tasks]
check-stow = { cmd = "which stow", description = "Verify stow is available (pixi-installed)" }
[target.osx-arm64.tasks]
check-stow = { cmd = "which stow || (echo 'Install with: brew install stow' && exit 1)", description = "Verify stow is available (brew-installed)" }
Key insight: Tasks with the same name in different [target.*.tasks] sections are automatically resolved based on the current platform. Target-agnostic tasks can depend on these platform-specific tasks by name.
Run tasks across multiple environments. Always test in all target environments before committing. Environment-specific failures in CI are expensive and avoidable.
[feature.py311.dependencies]
python = "3.11.*"
[feature.py312.dependencies]
python = "3.12.*"
[environments]
py311 = ["py311"]
py312 = ["py312"]
[tasks]
test = "pytest"
# Test in all Python versions
test-all = {
depends-on = [
{ task = "test", environment = "py311" },
{ task = "test", environment = "py312" },
]
}
[feature.build.dependencies]
rust = ">=1.70"
[feature.build.tasks]
compile = "cargo build --release"
[feature.test.dependencies]
pytest = "*"
[feature.test.tasks]
test-python = "pytest"
[environments]
build = ["build"]
test = ["test"]
[tasks]
ci = {
depends-on = [
{ task = "compile", environment = "build" },
{ task = "test-python", environment = "test" },
]
}
Use MiniJinja templates:
[tasks]
process = {
cmd = "python process.py inputs/{{ filename }}.txt",
args = ["filename"],
inputs = ["inputs/{{ filename }}.txt"],
outputs = ["outputs/{{ filename }}.out"]
}
pixi run process data1
[tasks]
lint = "ruff check ."
format = "ruff format --check ."
typecheck = "mypy src/"
unit-test = { cmd = "pytest tests/unit", depends-on = ["lint", "format"] }
int-test = { cmd = "pytest tests/integration", depends-on = ["unit-test"] }
build = { cmd = "python -m build", depends-on = ["int-test", "typecheck"] }
ci = { depends-on = ["build"] }
- uses: prefix-dev/setup-pixi@v0.9.2
with:
cache: true
- run: pixi run ci
YOU MUST:
NEVER: