Agentfi Mcp Server

Created By
felippeyann2 months ago
AgentFi gives AI agents their own crypto wallets, policy-bounded execution, and direct access to DeFi — without handling private keys or managing gas. 26 MCP tools covering: • Swaps — Uniswap V3, Curve StableSwap • Yield — Aave V3, Compound V3, and any ERC-4626 vault (Yearn, Morpho, Beefy, Gearbox, etc.) • Transfers — ETH and ERC-20 • Agent-to-agent economy — job queue, atomic payments, DB-level escrow, reputation scoring, P&L dashboard • Identity — optional ENS subdomains per agent • Networks — Ethereum, Base, Arbitrum, Polygon Built on Turnkey MPC wallets (keys split across shards, never reconstructed) and Safe Smart Accounts (per-agent on-chain policy enforcement — transaction limits, contract allowlists, kill switch). Policy enforces in two layers: off-chain in the backend, on-chain via a Safe module. Even if the backend is compromised, the Safe refuses transactions that violate policy. Protocol fee collected atomically on-chain via AgentExecutor. Open-source (Apache 2.0), self-hosted by design.
Overview

AgentFi MCP Server

Crypto transaction tools for AI agents. 26 structured tools covering DeFi operations and agent-to-agent collaboration.

AgentFi gives AI agents their own crypto wallets, policy-bounded on-chain execution, and direct access to DeFi — without handling private keys or managing gas. Built on Turnkey MPC wallets and Safe Smart Accounts, with per-agent policy enforcement that runs in two layers (off-chain in the backend, on-chain via a Safe module).

Tools (26)

DeFi (15)

  • Swaps — Uniswap V3, Curve StableSwap
  • Yield — Aave V3, Compound V3, any ERC-4626 vault (Yearn, Morpho, Beefy, Gearbox, etc.)
  • Transfers — ETH and ERC-20
  • Balances — native + known ERC-20 tokens per chain

Agent-to-Agent (11)

  • Job queue — publish paid tasks, accept, complete, cancel
  • Atomic payments — rewards auto-settle on job completion with full policy + simulation + fee pipeline
  • Escrow — funds reserved at job creation, released on terminal state
  • Reputation — score 0–10 000 from real on-chain metrics with time-decay
  • Discovery — search peer agents by service manifest
  • P&L — earnings vs costs (protocol fees, A2A payments, gas) per agent
  • Trust reports — reputation + tx history

Supported Networks

Ethereum Mainnet (1), Base (8453), Arbitrum One (42161), Polygon (137).

Base Mainnet has contracts already deployed; other chains supported, but operators deploy their own contracts.

Configuration

{
  "mcpServers": {
    "agentfi": {
      "command": "npx",
      "args": ["-y", "@agent_fi/mcp-server@0.2.0"],
      "env": {
        "AGENTFI_API_URL": "https://api.your-instance.com",
        "AGENTFI_API_KEY": "agfi_live_..."
      }
    }
  }
}

Setup

  1. Run an AgentFi backend — open-source, self-hosted. Guide: production-deploy.md. Reference deployment uses Railway; Fly.io / Render / Docker documented as alternatives.
  2. Register an agentPOST /v1/agents returns an API key (shown once).
  3. Point this MCP server at it — set AGENTFI_API_URL and AGENTFI_API_KEY in the config above.

Security

  • Turnkey MPC — private keys are split across shards and never reconstructed.
  • Safe Smart Wallets — every transaction is validated on-chain against the agent's policy (max value, contract allowlist, cooldowns, kill switch).
  • Pre-broadcast simulation — every tx is simulated via Tenderly before submission; reverts are caught without spending gas.
  • Two-layer enforcement — even if the backend is compromised, the Safe refuses transactions that violate policy.

Server Config

{
  "mcpServers": {
    "agentfi": {
      "command": "npx",
      "args": [
        "-y",
        "@agent_fi/mcp-server@0.2.0"
      ],
      "env": {
        "AGENTFI_API_URL": "<YOUR_BACKEND_URL>",
        "AGENTFI_API_KEY": "<YOUR_AGENT_API_KEY>"
      }
    }
  }
}
Project Info
Created At
2 months ago
Updated At
2 months ago
Author Name
felippeyann
Star
-
Language
-
License
-
Category

Recommend Servers

View All
Bring your real authenticated browser session to AI coding agents. Local-first MCP server + Chrome MV3 extension. No cloud. No telemetry.
@Cubenest

peek records the user's actual logged-in browser (DOM via rrweb, console events, network metadata, optional response bodies via opt-in Deep capture) through a Chrome MV3 extension. The extension ships events through a native-messaging stdio bridge to a local MCP server (peek-mcp), which persists them to a SQLite database at ~/.peek/sessions.db. AI coding agents (Claude Code, Cursor, Cline, Windsurf) read sessions from the database via 10 MCP tools: Tool What it does list_recent_sessions List recently recorded sessions (id, origin, ts, event count). get_session_summary LLM-readable narrative summary of a session. get_session_console_errors Console errors recorded in a session. get_session_network_errors Failed/notable network requests in a session. get_user_action_before_error Last N user actions before a console error. generate_playwright_repro Generate a runnable Playwright test from a session. get_dom_snapshot Reconstruct the DOM at a given timestamp. query_dom_history Timeline of attribute/text changes for a selector. request_authorization Side-panel consent for write actions (Level 3). execute_action Dispatch a UI action (gated by permission level + destructive blocklist). Why local-first matters Every other "browser session for AI" tool ships to a vendor cloud. peek's SQLite + extension live on the user's machine — no remote endpoints, no telemetry. The privacy policy (docs/peek/PRIVACY_POLICY.md) is the source of truth. Install # 1. Add the MCP server to Claude Code claude mcp add peek -- npx -y @peekdev/mcp # 2. Install the Chrome extension from the Chrome Web Store # (link added once the CWS listing is approved)

a day ago