TokRepo MCP Server

Name: TokRepo MCP Server
Author: henu-wang

@henu-wang

TokRepo MCP Server について

Search, browse, and install 500+ AI assets (agent skills, prompts, MCP configs, workflows) from any MCP client. Zero-dependency Node.js package.

#ai #mcp #skills #prompts #registry #claude-code #developer-tools

基本情報

カテゴリ

その他

ライセンス

MIT

ランタイム

node

トランスポート

stdio

公開者

henu-wang

投稿者

henu王凯

設定

以下の設定を使って、このサーバーを MCP 対応クライアントに追加してください。

{
  "mcpServers": {
    "tokrepo": {
      "command": "npx",
      "args": [
        "tokrepo-mcp-server"
      ]
    }
  }
}

ツール

Session bootstrap with high-trust assets, project memory pointer, recent handoffs, and policy pack URL

Planning-time capability discovery from a task, environment, and constraints

Atomic find action for a concrete task; returns ranked assets, match reasons, install command, and URL

Resolve a capability gap into a selected asset, verification evidence, install plan, lifecycle contract, next MCP calls, and CLI fallbacks

Search assets by keyword/tag with `agent_fit` ranking

Get full asset details by UUID

Inspect requires, extends, and co-used asset graph edges before planning installs

Get agent-native install plan v2 with rollback, evidence_bundle, SBOM-lite, and signature_evidence

Verify trust, hashes, permissions, policy, evidence_bundle, SBOM-lite, and signature_evidence before activation

Dry-run, stage, or install a Codex skill safely

List TokRepo-managed Codex installs

Dry-run or update managed Codex installs

Dry-run or remove a managed Codex install

Dry-run or roll back a prior Codex install session

Read-only packaging plan with quality_gate and package manifest for reusable local work after a task

Read-only package draft generator for reusable changed or explicit local files after a task

Push one explicit asset to TokRepo after user confirmation

概要

What is TokRepo MCP Server?

TokRepo MCP Server is an agent-native Model Context Protocol server that connects AI assistants to TokRepo, an open registry for reusable AI assets such as skills, prompts, MCP configs, scripts, and workflows. It enables agents to discover, verify, install, manage, and publish these assets through structured MCP tools.

How to use TokRepo MCP Server?

Install and connect using npx -y tokrepo-mcp-server. Add the server to your MCP client’s configuration (e.g., claude mcp add tokrepo, or in Cursor’s mcp.json). Once connected, the assistant gains access to over 20 tools for capability discovery, search, trust verification, install planning, lifecycle management, and post‑task harvesting.

Key features of TokRepo MCP Server

Agent‑native capability discovery and resolution from tasks
Search and browse public AI assets with agent‑fit signals
Trust verification with content hashes, permissions, and evidence bundles
Safe install planning with policy, rollback, and SBOM‑lite
Full lifecycle control: install, update, uninstall, rollback
Post‑task harvest and handoff planning for reusable local files
Project‑memory initialization across major agent platforms

Use cases of TokRepo MCP Server

An AI assistant discovers and installs a database MCP server during a planning step
A developer searches for trending AI assets by category (MCP, Skill, Prompt)
An agent verifies trust and install plan before activating a third‑party asset
A team harvests changed files after a task and publishes them as reusable assets
An agent rolls back a previous install after discovering compatibility issues

FAQ from TokRepo MCP Server

What is TokRepo?

TokRepo is an open registry for AI assets — like npm for packages, but for AI skills, prompts, MCP configs, and workflows. Assets are curated and include machine‑readable install contracts.

What are the requirements to run TokRepo MCP Server?

Node.js version 18 or higher and an internet connection to query the tokrepo.com API.

Does TokRepo collect telemetry?

Yes, it emits anonymous aggregate funnel events for capability resolution, discovery, verification, installs, harvests, and pushes. It does not send task text or file contents. Telemetry can be disabled with the environment variable TOKREPO_TELEMETRY=0.

How do I make future agents discover TokRepo?

Run npx tokrepo init-agent --target all once in a project. This writes managed instructions to multiple agent configuration files (e.g., AGENTS.md, CLAUDE.md, Cursor rules) so that agents know to call TokRepo during planning.