> ## Documentation Index
> Fetch the complete documentation index at: https://flox.dev/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# flox agent check-plugin

> Validate an agent plugin folder against the Agent Plugins and Agent Skills specifications.

## Synopsis

```text theme={null}
flox agent check-plugin <path> [--json] [--strict]
```

## Description

Validates a plugin folder against the [Agent Plugins specification](https://agent-plugins.org), and each of its skills against the [Agent Skills specification](https://agentskills.io).
It reports problems and never changes the folder.

The same check runs when `flox agent import` writes a package and when that package is built.

`<path>` is a plugin folder:

```text theme={null}
<plugin-name>/
├── plugin.json     # required manifest
├── skills/         # required, one folder per skill
│   └── <skill>/SKILL.md
└── mcp.json        # optional MCP server configuration
```

## Options

| Option | Effect |
| - | - |
| `--json` | Print the result as JSON: `valid`, `specVersion`, `errors`, and `warnings`, each issue with a `path` and a `message` |
| `--strict` | Treat warnings as errors |

## What is checked

* `plugin.json` exists, is valid JSON, and matches the schema its `$schema` names.
* The folder's name matches the manifest's `name`.
* `skills/` holds at least one skill: a folder directly under it with a file named exactly `SKILL.md`. Symbolic links are not followed.
* Each `SKILL.md` has frontmatter with a `name` and a `description` that follow the Agent Skills rules.
* `mcp.json`, when present, matches the MCP schema for the same specification version as `plugin.json`.
* Each local MCP server's `command` is a bare program name or a `./` path inside the plugin.

## Warnings and errors

A deviation that agents load anyway is a warning, and the plugin stays valid.
Common warnings are unknown frontmatter fields, such as `version:`, a skill `name` that does not match its folder, and a missing `description`.
Claude Code components such as `agents/` or `hooks/hooks.json` at the plugin root also warn, because packages keep them under `com.anthropic.claude-code/`.

Everything else is an error, including any schema violation and a symbolic link under `skills/`.

## Exit status

| Status | Meaning |
| - | - |
| `0` | Valid; warnings are allowed unless `--strict` is given |
| `1` | Invalid: one or more errors, or warnings under `--strict` |
| `2` | Usage error, or the path is missing or unreadable |

## Example

```bash theme={null}
flox agent check-plugin "$FLOX_ENV/share/agent-plugins/frontend-design"
```

## See also

* [`flox agent audit`](https://flox.dev/docs/ai-agents/reference/flox-agent-audit), which checks a plugin for security issues
* [Package a skill repository](https://flox.dev/docs/ai-agents/guides/package-a-skill-repository.md)


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.