score-studio-mcp
This server enables agents to identify users and organizations, inspect versioned datasets and model lineage, start training jobs, run and read evaluations, verify deployments, and execute typed workflows through authenticated MCP endpoints.
README
Score Studio agent plugins
Give your agents sight.
Score Studio is the computer vision layer for agents, automating data generation, annotation, training, evaluation, workflows, and deployment in one evidence-backed system.
This repository brings that complete vision loop to the major agent ecosystems through one portable MCP server, without publishing private source identifiers or internal implementation details.
What an agent can do
- identify the signed-in user and organization memberships;
- inspect versioned datasets and model lineage;
- start tracked training jobs with explicit compute and dataset versions;
- list measured evaluations and read immutable reports;
- start evaluations against exact model/dataset version IDs;
- inspect and verify deployments;
- list and run typed workflows while reporting deferred blocks truthfully.
The MCP tools are intentionally narrower than the full Score Studio API. Delete, billing, member administration, provider credentials, and release switching are not exposed in the first version.
Local server
Requires Node.js 20+ and a Score Studio bearer token or API key.
npm install
npm run build
SCORESTUDIO_TOKEN=your-token npm start
The API defaults to https://api.scorestudio.ai. Set SCORESTUDIO_URL only to
target another Score Studio environment.
For local Streamable HTTP development:
npm run start:http
The endpoint is http://127.0.0.1:3000/mcp; each request must include its own
Authorization: Bearer … header. /healthz does not require authentication.
Container deployment
The production image listens on 0.0.0.0:3000 and uses the production Score
Studio API by default; callers provide their own bearer token on each request.
docker build -t score-studio-mcp .
docker run --rm -p 3000:3000 score-studio-mcp
Packages
| Host | Package | Discovery route |
|---|---|---|
| ChatGPT + Codex | plugins/openai/score-studio |
OpenAI universal plugin directory |
| Claude Code | plugins/claude/score-studio |
Anthropic marketplace or this repo's marketplace |
| Gemini CLI | plugins/gemini/score-studio |
Git install and Gemini extension gallery |
| Cursor | Existing score-studio-cursor-plugin |
Cursor Marketplace |
| Lovable | Shared remote MCP endpoint | Custom chat connector; no equivalent public plugin submission flow yet |
| GitHub Copilot / VS Code | Shared local or remote MCP endpoint | Manual MCP config, then GitHub MCP Registry when eligible |
| Windsurf | Shared local or remote MCP endpoint | MCP Marketplace submission or manual Cascade config |
| Other MCP hosts | Shared remote MCP endpoint | Streamable HTTP /mcp |
See distribution.md for installation, deployment, and marketplace requirements.
Authentication and release boundary
Local packages read SCORESTUDIO_TOKEN from the host's secret configuration
and optionally accept a SCORESTUDIO_URL override. They do not persist or print
secrets.
A public ChatGPT listing cannot use this manual-token flow. It requires a stable public HTTPS MCP endpoint plus OAuth 2.1 discovery/authorization. That hosted OAuth boundary is deliberately left as a release gate; the repository does not claim the package is publicly installable before it exists.
License
Apache License 2.0.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.