mcpField
T
Trustbench

Trustbench

Remote
github.com/lithvallFinance0Updated 1d ago
Repository

About Trustbench

Trustbench is a read-only MCP server for x402 agent payments, offering provider rankings, receipt lookup, and verification. It checks Ed25519 signatures and on-chain status, and can verify receipt JSON offline.

Key features

  1. 01Rank x402 providers by capability
  2. 02Fetch receipts by ID
  3. 03Verify Ed25519 signatures
  4. 04Check on-chain receipt status
  5. 05Verify receipt JSON offline

Use cases

  1. 01Compare x402 providers by capability
  2. 02Inspect a payment receipt by ID
  3. 03Check a receipt's signature and chain status
  4. 04Verify a receipt received from a third party

MCP support

It exposes three read-only MCP tools over stdio: get_rankings, get_receipt, and verify_receipt.

Configuration

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

Endpointhttps://trustbench.io/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.

Packages
npm@trustbench/mcp1.1.1

Tools

3
ParameterTypeRequiredDescription
capabilityenumYesThe provider capability to query rankings for.search | inference | data | media | infra

Details

From the project README

@trustbench/mcp

Native MCP server for TrustBench — a non-custodial routing and audit layer for x402 agent payments.

Adds three read-only tools to any MCP-capable host:

ToolWhat it does
get_rankingsScored x402 provider list by capability. No API key.
get_receiptFetch any TrustBench receipt by ID (rcpt_… / rrcpt_…). No API key.
verify_receiptConfirm Ed25519 signature + on-chain status. Two modes: lookup by ID, or offline verification of a receipt JSON. No API key.

Works in Claude Desktop, Claude Cowork, Grok (xAI Connectors), Kimi Code CLI, Cherry Studio, Cursor, and any MCP-compatible host.

Setup

Add to your host's MCP config file (e.g. claude_desktop_config.json for Claude Desktop, ~/.kimi/mcp.json for Kimi):

{
  "mcpServers": {
    "trustbench": {
      "command": "npx",
      "args": ["-y", "@trustbench/mcp"]
    }
  }
}

Restart the host app. The three tools above become available immediately — no API key, no account required.

Environment variables

VariableDefaultDescription
TRUSTBENCH_BASE_URLhttps://trustbench.ioOverride the API base (useful for local dev).

What the tools return

get_rankings — JSON array of scored providers for the requested capability. Capabilities: search, inference, data, media, infra. Scores are derived from HEAD-probe liveness checks (3 samples from one host). See trustbench.io/methodology for the full scoring formula.

get_receipt — Full signed receipt envelope. Receipts are immutable after issuance. Phase 3 IDs start with rcpt_, Phase 4 routing receipts start with rrcpt_.

verify_receipt — Two modes (provide exactly one input):

  • Lookup mode (receipt_id) — fetches the receipt from trustbench.io and re-runs Ed25519 + on-chain verification. Useful when the agent only has an ID.
  • Offline mode (receipt_json) — verifies a full {receipt, signature} envelope an agent received from a third party against the published public key, without trusting the database. The third-party-verifier path.

Both modes return a JSON summary with signature_valid (boolean), on_chain_verified (boolean where applicable), signature_alg (ed25519), and pubkey_url. For zero-network verification with no round-trip to TrustBench, use the @trustbench/verify-receipt npm package instead.

Routing tools (v1.5)

route_quote and route_settle (the paid routing flow) require a tb_live_… API key and ship in the next release. Request access by DM'ing @TrustBench on X.

Technical details

  • Transport: stdio (JSON-RPC 2.0) — the MCP standard
  • No external dependencies beyond Node.js built-ins
  • Node ≥ 18 required (uses native fetch and readline)
  • Source: github.com/lithvall/TrustBench

Related packages

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.