@matihlabs/mcp

Connect AI clients (Claude Desktop, ChatGPT, Cursor, …) to the Matih data platform over the Model Context Protocol. Write SQL, profile tables, run analyses + charts, manage dashboards, and upload files — all through MCP tools, with OAuth/PKCE auth and PII-safe egress.

Use as a stdio MCP server (Claude Desktop / Cursor)

Add to your client's MCP config. The token is delivered via the environment — never as a CLI flag (which leaks into the OS process list):

{
  "mcpServers": {
    "matih": {
      "command": "npx",
      "args": ["-y", "@matihlabs/mcp", "https://app.matih.ai/api/v1/mcp"],
      "env": { "MATIH_MCP_TOKEN": "<your Matih bearer token>" }
    }
  }
}

The bridge turns the remote Matih HTTP MCP endpoint into a local stdio MCP server, forwarding every tools/call, resources/read, and prompts/get to Matih.

Use as a library

import { McpClient, MatihTools, StaticTokenProvider } from "@matihlabs/mcp";

const client = new McpClient({
  endpoint: "https://app.matih.ai/api/v1/mcp",
  tokenProvider: new StaticTokenProvider(process.env.MATIH_MCP_TOKEN!),
});
await client.initialize();

const matih = new MatihTools(client);
const result = await matih.runSql({ connection_id: "<id>", sql: "select 1" });

OAuth (PKCE) instead of a static token

import { McpClient, OAuthTokenProvider } from "@matihlabs/mcp";

const tokenProvider = new OAuthTokenProvider({
  resourceMetadataUrl: "https://app.matih.ai/.well-known/oauth-protected-resource/api/v1/mcp",
  clientId: "<registered client id>",
  acquire: async ({ metadata, clientId, resource }) => {
    // open metadata.authorization_endpoint (PKCE S256, resource=<resource>),
    // capture the code at your redirect_uri, return { code, verifier, redirectUri }.
  },
});

The provider runs RFC 9728 → RFC 8414 discovery, PKCE S256, RFC 8707 resource-bound tokens, caches, and refreshes; a 401 invalid_token triggers one re-auth.

Tools

run_sql, run_analysis, profile_table, create_chart, create_dashboard, get_dashboard, publish_dashboard, whoami, get_scope, get_query_result, upload_file, upload_status, create_upload_url, finalize_upload (+ catalog / lineage resources and the explain_metric prompt).

Notes

• Egress consent. Matih gates third-party-LLM data egress per tenant. If your tenant hasn't accepted the data-processing agreement, calls return a clear EGRESS_CONSENT_REQUIRED error with a link to accept it.
• Node ≥ 20 (uses native fetch + node:crypto; zero runtime dependencies).

License: Apache-2.0 · https://matih.ai