Skip to main content
MCP server CX Codex

MolTrust — Claude Code MCP server: decentralized identity and agent reputation

MolTrust — an MCP server that registers AI agents on the Base blockchain, issues W3C Verifiable Credentials, and scores wallet trust from 0 to 100. Runs in Claude Code, Codex, Cursor, GitHub Copilot, Windsurf, Cline, Zed, and Gemini CLI.

no license 1,453 installs updated 2026-03-01

Install into Claude Code · Platform CLI: remote server All methods and platforms ↓
claude mcp add --transport http moltrust-mcp-server https://moltrust-mcp-server--moltrust.run.tools
Open on GitHub

What MolTrust does

MolTrust connects an AI agent to a decentralized trust network on Base (Ethereum L2). Calling moltrust_register creates a DID, anchors it via ERC-8004, and issues a W3C Verifiable Credential. Any counterpart agent can then call moltrust_verify to retrieve that trust card — reputation score, credentials, and blockchain anchor — before engaging in a peer-to-peer transaction.

Credits are purchased with USDC on Base at a fixed rate of 1 USDC = 100 credits. After sending USDC to the MolTrust wallet, you submit the transaction hash through moltrust_claim_deposit along with your agent's DID to receive credits on-chain. The MoltGuard toolset scores a Base wallet from 0 to 100 by examining on-chain activity, ERC-8004 registration, and counterparty diversity. It also scans for Sybil rings by tracing ETH funding clusters and checks Polymarket markets for volume spikes or price-volume divergence.

For agents that make purchases or book travel on behalf of a human, the server issues scoped Verifiable Credentials (mt_shopping_issue_vc, mt_travel_issue_vc) with cryptographically enforced spend limits and segment permissions. The mt_skill_audit tool fetches a SKILL.md from a URL, computes its SHA-256 hash, and runs an eight-point security audit covering prompt injection, data exfiltration, and tool scope violations — a missing SKILL.md will cause the audit to fail.

What it helps with

  • Register an AI agent in a decentralized registry with a DID and Verifiable Credential
  • Check another agent's reputation score before a peer-to-peer transaction
  • Detect Sybil wallets by analyzing transaction history and ETH funding clusters
  • Issue a BuyerAgentCredential with enforced spend limits for a shopping agent
  • Audit an agent's SKILL.md for prompt injection and data exfiltration risks

Example prompts for the agent

  • Register a new agent named my-agent on platform openai using moltrust_register
  • Verify the trust card for did:moltrust:a1b2c3d4e5f60718 with moltrust_verify
  • Scan wallet 0x... for Sybil indicators using moltguard_sybil

How to install MolTrust

Platform CLI: remote server

claude mcp add --transport http moltrust-mcp-server https://moltrust-mcp-server--moltrust.run.tools

Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more

Questions about MolTrust

How do I install MolTrust in Claude Code?

Run: claude mcp add --transport http moltrust https://moltrust-mcp-server--moltrust.run.tools

What does MolTrust do?

It registers AI agents on the Base blockchain with DIDs and W3C Verifiable Credentials, tracks reputation ratings from 1 to 5, manages credit balances funded by USDC, and detects Sybil wallets along with integrity anomalies on Polymarket markets.

What does MoltGuard's trust score measure?

The score ranges from 0 to 100 and factors in on-chain activity, ERC-8004 registration status, USDC balance, counterparty diversity, and MolTrust credentials tied to the Base wallet address.

Similar MCP servers

All MCP servers for Agent setup: memory, context, orchestration →

Project sponsors

Companies that keep this analytics open