TensorBlock MCP Index API

The TensorBlock MCP Index API exposes the generated catalog as HTTP JSON. It is intended for agents, clients, and lightweight integrations that need to search MCP servers or generate install configs without parsing markdown.

Hosted API:

https://mcp-index.tensorblock.co

Endpoints

GET /
GET /health
GET /v1
GET /v1/categories
GET /v1/servers?query=&category=&transport=&auth=&limit=
GET /v1/servers/recent?limit=
GET /v1/servers/updated?limit=
GET /v1/servers/{id}
GET /v1/servers/{id}/badge.svg
GET /v1/servers/{id}/install-config?client=claude-desktop|cursor|codex|vscode

Examples

Discover available endpoints:

curl "$MCP_INDEX_API_URL/"

Check API health and deployment fingerprint:

curl "$MCP_INDEX_API_URL/health"

{
  "status": "ok",
  "catalogEntries": 7729,
  "loadedAt": "2026-07-08T00:00:00.000Z",
  "build": {
    "commitSha": "694bc1a8c555e6d6df6dfe2ee1b04b7ae866eee5",
    "builtAt": "2026-07-08T00:11:55.000Z"
  },
  "uptimeSeconds": 60
}

Search for Postgres servers:

curl "$MCP_INDEX_API_URL/v1/servers?query=postgres&limit=5"

Filter by category and transport:

curl "$MCP_INDEX_API_URL/v1/servers?category=Databases&transport=stdio"

Fetch recently added servers:

curl "$MCP_INDEX_API_URL/v1/servers/recent?limit=12"

Fetch recently updated catalog entries:

curl "$MCP_INDEX_API_URL/v1/servers/updated?limit=12"

Server summaries include sourcePullRequest when known and lastUpdatedAt when the catalog builder can derive it from git history. The recent and updated collections sort by lastUpdatedAt, using source pull request numbers as an additional ordering signal when timestamps tie.

Fetch a full server profile:

curl "$MCP_INDEX_API_URL/v1/servers/github-crystaldba-postgres-mcp-22e80ea8"

Open a shareable website profile page:

https://tensorblock.co/mcp/servers/github-crystaldba-postgres-mcp-22e80ea8

The API also keeps GET /servers/{id} as a lightweight HTML fallback, but the canonical community profile is hosted on the TensorBlock website.

Embed an MCP Index badge in a project README:

[![Indexed on TensorBlock MCP Index](https://mcp-index.tensorblock.co/v1/servers/github-crystaldba-postgres-mcp-22e80ea8/badge.svg)](https://tensorblock.co/mcp/servers/github-crystaldba-postgres-mcp-22e80ea8)

Generate an install config:

curl "$MCP_INDEX_API_URL/v1/servers/github-crystaldba-postgres-mcp-22e80ea8/install-config?client=claude-desktop"

Railway Deployment

The API is designed to run as a Railway service from this repository.

Build command:

npm run catalog:build && npm run profiles:build && npm run build

Start command:

npm run start

The service reads data/catalog.json on startup and keeps it in memory. Every deploy includes the latest generated catalog from the repository. No database is required for the MVP.

Automatic Refresh

The deployment workflow runs on every push to main that touches the catalog, docs, API package, schema, or Railway config. It rebuilds data/catalog.json and data/profiles/*.json, writes data/build-info.json, and verifies the live /health deployment fingerprint before finishing, so newly merged server entries are reflected by the live API after the Railway deployment succeeds.

The workflow requires a GitHub repository secret named RAILWAY_TOKEN with deploy access to the Railway project.