mcpField
L
Lemma — Verifiable Provenance For Ai Agents

Lemma — Verifiable Provenance For Ai Agents

Remote
lemma.frame00.comSecurity1Updated today

About Lemma — Verifiable Provenance For Ai Agents

Lemma’s MCP server lets AI agents query cryptographically verified attributes derived from confidential documents without exposing their plaintext. It is intended for applications that need selective disclosure and tamper-evident provenance; the current release provides read-only tools.

Key features

  1. 01Query verified document attributes
  2. 02Retrieve schema definitions
  3. 03Retrieve proof circuits
  4. 04Retrieve document generators
  5. 05Check proof verification status

Use cases

  1. 01Check verified attributes without revealing documents
  2. 02Inspect an attribute schema
  3. 03Review a proof circuit or generator
  4. 04Check the status of a submitted proof

MCP support

The server exposes five read-only MCP tools that query Lemma’s API for verified attributes, schemas, circuits, generators, and proof status. Resources and prompts are not included in the current MVP.

Configuration

Remote servers need no local install. Pick your client, then copy the config or install in one click.

Endpointhttps://mcp.lemma.workers.dev/mcp
Copy install command
Claude CodeTerminal
Cursor.cursor/mcp.json
VS Code.vscode/mcp.json
Claude Desktopclaude_desktop_config.json
Windsurf~/.codeium/windsurf/mcp_config.json

Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.

All endpoints
Streamable HTTPhttps://mcp.lemma.workers.dev/mcpAuth required
Packages
npm@lemmaoracle/mcp0.0.17

Tools

5
ParameterTypeRequiredDescription
limitnumberNoMax results per page (1–200). Defaults to API server default (50).
offsetnumberNoPagination offset. Pair with `hasMore` in the response to walk pages.
schemasarrayNoRestrict results to documents conforming to these schema IDs.
chainIdsarrayNoRestrict results to attributes verified on these chain IDs (EVM).
attributesarrayNoAttribute predicates to AND-combine.

Details

From the project README

@lemmaoracle/mcp

LightNow capabilities npm version

@lemmaoracle/mcp is the Model Context Protocol (MCP) server that gives AI agents access to Lemma's verifiable provenance layer — query cryptographically verified attributes (zero-knowledge proofs, selective disclosure, tamper-evident provenance) from confidential documents without ever exposing the plaintext.

Models change. Proofs remain.

What is this?

This is an MCP server. MCP (Model Context Protocol) is the open protocol for connecting AI agents to external tools and data sources, originally released by Anthropic and now maintained as a community standard. This package implements an MCP server using the official MCP SDK (@modelcontextprotocol/sdk) and exposes a set of read-only tools for querying Lemma's verifiable attribute API.

The current MVP exposes 5 read tools. Resources and prompts are not exposed in v0.0.x; write tools (register_document, submit_proof) are planned for Phase 2.

Quick start

Two ways to connect — pick whichever matches your agent setup.

Option 1 — Smithery (hosted gateway, zero install)
npx -y smithery mcp add lemmaoracle/lemma

Smithery proxies to the Lemma MCP endpoint at https://lemma--lemmaoracle.run.tools and prompts for your Lemma API key. Best for trying the server quickly or for agents that already speak Smithery. Server page: smithery.ai/servers/lemmaoracle/lemma.

Option 2 — Claude Desktop with npx (stdio)

Add this to your Claude Desktop MCP config:

{
  "mcpServers": {
    "lemma": {
      "command": "npx",
      "args": ["-y", "@lemmaoracle/mcp"],
      "env": {
        "LEMMA_API_KEY": "YOUR_API_KEY"
      }
    }
  }
}

Get an API key at https://lemma.frame00.com/services. LEMMA_API_BASE can be omitted (defaults to production endpoint).

Try it in Glama Inspector

When trying this server in Glama's in-browser inspector (or any MCP inspector), you can use the demo key below. It is read-only and scoped to demo data (the same key is used by example-x402):

LEMMA_API_KEY=b6363aa6265322ed0d786a11d5b6d3264947052ca72deba4cbe1685d099af892
LEMMA_API_BASE=https://workers.lemma.workers.dev

For production access, request a key at https://lemma.frame00.com/services.

Tools (MCP tool list)

This server exposes the following MCP tools (no resources or prompts in the current MVP):

ToolPhaseDescription
lemma_query_verified_attributesMVPQuery cryptographically verified attributes from Lemma. Use this as the primary entry point for finding documents whose attributes match given conditions.
lemma_get_schemaMVPRetrieve a Lemma schema (attribute structure) by ID.
lemma_get_circuitMVPRetrieve a ZK proof circuit (constraints + verifier) by ID.
lemma_get_generatorMVPRetrieve a document generator (input/output spec) by ID.
lemma_get_proof_statusMVPGet the verification status of a proof by its verificationId.
lemma_register_documentPhase 2Register a new document on Lemma.
lemma_submit_proofPhase 2Submit a ZK proof for verification.

Architecture

  • MCP Server: this package — connects AI agents to the Lemma API via the Model Context Protocol.
  • Lemma API: REST surface deployed on Cloudflare Workers, defined by OpenAPI v2.
  • MCP SDK: this server is built on the official @modelcontextprotocol/sdk (≥ 1.29).
  • License: Apache-2.0.

Privacy & data handling

Lemma is designed so that we never receive plaintext documents. The MCP server only transmits commitments, hashes, and ZK proofs to the Lemma API. See https://lemma.frame00.com/privacy for the full privacy policy.

Local development (contributors)

For contributors hacking on the MCP server itself, run it from a local build of this monorepo. End users should use the npx config above.

git clone https://github.com/lemmaoracle/lemma.git
cd lemma
pnpm install
pnpm -F @lemmaoracle/mcp build

The bin entry produces packages/mcp/dist/index.js. Point Claude Desktop at the absolute path:

{
  "mcpServers": {
    "lemma": {
      "command": "node",
      "args": ["/absolute/path/to/lemma/packages/mcp/dist/index.js"],
      "env": {
        "LEMMA_API_KEY": "YOUR_API_KEY"
      }
    }
  }
}

Rebuild after changes with pnpm -F @lemmaoracle/mcp build and restart Claude Desktop to reconnect. To point at a non-production API, add "LEMMA_API_BASE": "https://..." to env.

Resources


🇯🇵 日本語版はこちら / Japanese README

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.