Linux command
vaen 命令
文本
复制后可按需替换文件名、目录或参数。
常用示例
Validate
vaen validate
Build
vaen build -f agent.yaml -o my-setup.agent
Inspect
vaen inspect my-setup.agent
Import
vaen import my-setup.agent --client claude --target /path/to/repo
Doctor
vaen doctor --client claude --target /path/to/repo
Clean up
vaen cleanup
说明
vaen is a Python CLI that packages an AI coding-agent setup (instructions, skills and MCP server declarations) into a portable, OCI-style archive with a `.agent` extension, so it can be shared across repositories or teammates without copy-pasting files. A `.agent` bundle declares which instruction files exist, which skills it bundles, which MCP servers it expects, and which environment-variable names hold credentials. It never contains credential values, only the names of variables the receiver must populate locally. `import` materializes the bundle into a target repo for a specific client (for example Claude Code) and writes files to the expected locations. `doctor` verifies that an import is structurally valid. vaen exists because a zip file can move files but cannot describe what the setup is, where files should land, or which secrets the receiver must provide.
参数
- validate
- Check the manifest syntax.
- build
- Build a `.agent` archive. Flags: `-f` manifest, `-o` output filename.
- inspect _archive_
- Print the contents of a `.agent` archive without importing.
- import _archive_
- Materialize the bundle into a target. Flags: `--client`, `--target`, `--target-instructions-file-name`, `--target-skills-directory`.
- doctor
- Validate an imported structure. Accepts the same client and target flags as `import`.
- cleanup
- Remove the locally stored canonical copy.
FAQ
What is the vaen command used for?
vaen is a Python CLI that packages an AI coding-agent setup (instructions, skills and MCP server declarations) into a portable, OCI-style archive with a `.agent` extension, so it can be shared across repositories or teammates without copy-pasting files. A `.agent` bundle declares which instruction files exist, which skills it bundles, which MCP servers it expects, and which environment-variable names hold credentials. It never contains credential values, only the names of variables the receiver must populate locally. `import` materializes the bundle into a target repo for a specific client (for example Claude Code) and writes files to the expected locations. `doctor` verifies that an import is structurally valid. vaen exists because a zip file can move files but cannot describe what the setup is, where files should land, or which secrets the receiver must provide.
How do I run a basic vaen example?
Run `vaen validate` in a terminal, then adjust file names, paths, flags, or remote targets for your system.
What does validate do in vaen?
Check the manifest syntax.