# dossier [Health: Active]

**Category:** 💻 Developer Tools  
**Repository:** https://github.com/imboard-ai/ai-dossier  
**GitHub Stars:** 10  
**npm Downloads (last month):** 20555  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/dossier

## Description
MCP server for dossier automation standard - enables LLMs to discover, verify, and execute dossiers

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "dossier": {
    "command": "npx",
    "args": ["-y","@ai-dossier/cli"]
  }
}
```

## Documentation

## What dossier MCP server does

dossier MCP server brings the Dossier skill format into MCP-enabled AI clients. It lets an agent find available dossiers, check a dossier’s integrity and authenticity, and run a selected workflow. Dossiers use `.ds.md` files with metadata, instructions, a version, and a signature. The format is designed for skills that can be shared across tools while preserving the same pinned content.

Dossier is intended for model-driven orchestration. The model reads a skill and decides which steps and tool calls are needed, while Dossier provides a deterministic layer around the run. This differs from a hand-coded workflow engine: the instructions guide the model inside the skill, and verification happens before the skill is executed.

## How it works

A signed dossier combines its skill instructions with a checksum and Ed25519 signature. Verification checks that the content has not changed and confirms the signer associated with the dossier. Version pinning helps different users or machines install and run the same skill version.

Skills can be hosted in a GitHub repository, on another server, or through the Dossier registry. Their location does not change the verification model because the version and signature travel with the skill. Dossier also works as a layer over Agent Skills and `SKILL.md`: skills can be moved into and out of the format with the Dossier CLI.

Once the MCP server is configured, an agent can request available dossiers, verify one, and execute it without manually copying a dossier URL into the conversation. The server is distributed as `@ai-dossier/mcp-server` and can be launched with `npx`.

## Setup and configuration

For Claude Code, register the server with:

```bash
claude mcp add dossier --scope user -- npx @ai-dossier/mcp-server
```

A generic MCP client can use the same stdio command. The equivalent configuration uses `npx` as the command and passes `-y` and `@ai-dossier/mcp-server` as arguments:

```json
{
  "mcpServers": {
    "dossier": {
      "command": "npx",
      "args": ["-y", "@ai-dossier/mcp-server"]
    }
  }
}
```

After setup, prompts such as “List available dossiers” or a request to run a named dossier can invoke the server’s capabilities. The repository also provides a Claude Code plugin installation path.

## Tools and capabilities

- Discover available dossiers for an agent to inspect.
- Verify dossier checksums and signatures before execution.
- Execute a selected dossier through the MCP connection.
- Work with signed, versioned `.ds.md` skills.
- Support dossiers hosted in Git repositories, other servers, or the Dossier registry.
- Use the Dossier workflow alongside Agent Skills and `SKILL.md` files.

The signatures establish integrity and origin; they do not make a dossier safe by themselves or block prompt injection. The skill’s instructions still need appropriate review before use.

_Full upstream README: https://allmcps.com/mcp/dossier/readme_