@ai-dossier/mcp-server weekly downloads statistics

@ai-dossier/mcp-server

Use one portable AI-agent skill across Claude Code, Codex, OpenCode, and MCP-compatible workflows.

@ai-dossier/mcp-server is for teams that want their MCP clients to discover, verify, and run trusted skills through the Model Context Protocol. Dossiers are portable, signed, versioned skills: familiar to Claude Code users, but verifiable and registry-backed for any compatible LLM tool.

Part of the ai-dossier ecosystem: use @ai-dossier/cli to author and install skills, @ai-dossier/core for verification, and the ai-dossier project for the complete platform.

Installation

Claude Code Plugin (Recommended)

/plugin marketplace add imboard-ai/ai-dossier
/plugin install dossier-mcp-server@ai-dossier

One-time setup — auto-updates included.

Claude Code (Manual)

# Global (available across all projects)
claude mcp add dossier --scope user -- npx @ai-dossier/mcp-server

# Project-only
claude mcp add dossier -- npx @ai-dossier/mcp-server

Claude Desktop

Add to claude_desktop_config.json:

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

Verify

claude mcp list

Then try:

"List available dossiers in this project"
"Execute the project-init dossier"

---

Tools

Discovery & Reading

Tool	Description
list_dossiers	Discover available dossiers in a directory. Scans for *.ds.md files and returns metadata.
read_dossier	Read and parse a dossier file. Returns metadata and content. Should be called after verify_dossier passes. Pass dry_run: true to also get dry_run_plan (static preview).
search_dossiers	Search the dossier registry for available dossiers by keyword and optional category filter.

Security Verification

Tool	Description
verify_dossier	Verify integrity (checksum) and authenticity (signature) of a single dossier. Returns pass/fail with stage details.
verify_graph	Batch verify all dossiers in a resolved dependency graph. Returns aggregate security report with per-dossier breakdown.

Orchestration (Multi-Dossier Journeys)

Tool	Description
resolve_graph	Resolve a dossier dependency graph into an execution plan. Produces a DAG with ordered phases, parallel groups, and conflict detection.
start_journey	Start a journey session from a resolved and verified graph. Returns the first step's dossier content with injected context. With dry_run: true no session is created and a static per-step preview (files, commands, network, env, risk score) is returned instead. The preview is derived from declared metadata and code blocks; the executing agent may take other actions and the score is a heuristic, not a safety guarantee.
step_complete	Mark the current journey step as complete or failed. Advances to the next step with context from previous outputs.
get_journey_status	Get the current state of a journey: completed steps, current step, remaining steps, and collected outputs.
cancel_journey	Cancel an active journey session. Returns a summary of what completed before cancellation.

---

Resources

URI	Description
dossier://concept	Introduction to dossiers — what they are and why to use them
dossier://protocol	How to execute dossiers safely and effectively
dossier://security	Security model, signing, verification, and trust management
dossier://orchestration	Complete reference for multi-dossier journey tools

---

Prompts

Prompt	Description
execute-dossier	Run a dossier with verification and protocol. Automatically chooses single or multi-dossier flow based on relationships.
execute-journey	Guide through a multi-step dossier journey: resolve, verify, present plan, execute steps, collect outputs.
create-dossier	Author a new dossier using the official meta-dossier template.

---

Security

Every dossier execution is verified before running:

User: "Execute the deploy-to-aws dossier"

LLM:  *Calls verify_dossier*

      Security Verification:
      - Integrity: Checksum verified (content not tampered with)
      - Authenticity: Signed by imboard-ai-2024 (trusted)
      - Risk Level: HIGH

      This dossier will modify AWS infrastructure.
      Proceed? (y/N)

Verification results

• ALLOW — Verified signature from trusted source + low risk. Execute confidently.
• WARN — Unsigned/unknown signer or high risk. Request user approval.
• BLOCK — Checksum failed or signature invalid. Do not execute.

Features

• Integrity: SHA256 checksums ensure dossiers haven't been modified
• Signatures: Optional Ed25519 or AWS KMS cryptographic signatures (legacy minisign keys still verify)
• Risk assessment: Declared risk level and specific risk factors per dossier
• Trust model: Decentralized — users choose which signing keys to trust via ~/.dossier/trusted-keys.txt

---

Development

git clone https://github.com/imboard-ai/ai-dossier.git
cd ai-dossier/mcp-server
npm install
npm run build
npm test
npm start

Project structure…