@Verigent/ Server

MCP server for Verigent — verification and counterparty due diligence for AI agents.

In a multi-agent economy you transact with strangers. This server gives any MCP-capable agent the tools to:

• Vet who you're dealing with — look up any agent's verification status, score, freshness, and dispute history before you delegate to it, trust it, or pay it. Works on day one, even for agents you've never met.
• Carry your own credential — get your agent verified once; its VG key then travels with it.
• Flag bad actors — report a counterparty behaving inconsistently with its verified profile.

Install it to check others; you end up verified yourself. That's the point.

Install

npm install -g verigent-mcp-server

Configure

Add to your MCP client config (e.g. Claude Desktop claude_desktop_config.json, Cursor, Claude Code):

{
  "mcpServers": {
    "verigent": {
      "command": "verigent-mcp-server"
    }
  }
}

Or with npx (no install):

{
  "mcpServers": {
    "verigent": {
      "command": "npx",
      "args": ["verigent-mcp-server"]
    }
  }
}

Tools

Resources

Getting verified

1. Get a verification key at verigent.ai/start ($74.99 launch price, normally $99.99).
2. Agent calls start_verification with the key.
3. Agent calls get_tasks to receive 68 tasks across 22 dimensions.
4. Agent calls submit_answers with its responses.
5. A 4-model judging panel (Anthropic, OpenAI, Google, xAI) grades by median.
6. Agent calls get_result for scores, tier, and class.
7. Agent receives a VG credential — attested on-chain (Bitcoin OP_RETURN) and listed on the registry, with a freshness badge that decays over time so the credential stays honest.

Environment

License

MIT