Arcadia Finance MCP Server

Arcadia Finance MCP Server

MCP server for Arcadia Finance, a platform to manage concentrated liquidity positions on Uniswap and Aerodrome with automated rebalancing, compounding, yield optimization, and leverage. Read protocol data and build unsigned transactions for LP management, borrowing, deposits, and more.

Designed for AI agents (Claude, Cursor, etc.) to interact with Arcadia onchain.

Install

Tools

Read Tools

Write Tools

All write tools return unsigned transactions as { to, data, value, chainId }.

Dev Tools

Always registered but requires PK env var to function.

Transaction Signing

All write tools return unsigned transactions as { to, data, value, chainId }. This server does NOT sign or broadcast — your agent or application is responsible for that.

Options

Wallet MCP servers (recommended for production): Pair this server with a wallet MCP server that handles signing:

Or use your existing wallet setup (Fireblocks, Dfns, Turnkey, Biconomy, Dynamic) and pass the unsigned tx object to your provider's signing method.

viem/ethers in your agent:

import { createWalletClient, http } from "viem";
import { privateKeyToAccount } from "viem/accounts";
import { base } from "viem/chains";

const account = privateKeyToAccount("0x...");
const client = createWalletClient({ account, chain: base, transport: http() });

// tx = result from any write.* tool
const hash = await client.sendTransaction(tx);

Built-in dev.send tool (development only): The server includes a dev-only signing tool that reads a private key from the PK environment variable. Set PK via a .env file or your MCP client config:

# .env in the server directory (never commit — already gitignored)
PK=0xYourPrivateKeyHex
RPC_URL_BASE=https://base-mainnet.g.alchemy.com/v2/your-key

The server loads .env automatically on startup. Works with any MCP client (Claude Desktop, Claude Code, VSCode, Cursor). MCP client env block settings take precedence if both are set.

Not for production — use a dedicated wallet MCP server (Fireblocks, Turnkey, Safe) instead.

Setup

Prerequisites: Node.js >= 22

yarn install
yarn build

Environment variables:

Supported chains: Base (8453), Unichain (130)

MCP Client Configuration

Remote (no install needed):

{
  "mcpServers": {
    "arcadia-finance": {
      "url": "https://mcp.arcadia.finance/mcp"
    }
  }
}

Via npx (local stdio):

{
  "mcpServers": {
    "arcadia-finance": {
      "command": "npx",
      "args": ["-y", "@arcadia-finance/mcp-server"],
      "env": {
        "RPC_URL_BASE": "https://base-mainnet.g.alchemy.com/v2/your-key"
      }
    }
  }
}

Claude Code:

# Remote
claude mcp add arcadia-finance --transport http https://mcp.arcadia.finance/mcp

# Local
claude mcp add arcadia-finance -- npx -y @arcadia-finance/mcp-server

From source (local development):

{
  "mcpServers": {
    "arcadia-finance": {
      "command": "node",
      "args": ["/absolute/path/to/mcp-server/dist/index.js"],
      "env": {
        "RPC_URL_BASE": "https://base-mainnet.g.alchemy.com/v2/your-key"
      }
    }
  }
}

Skills

The skills/ directory contains Claude Code skills that teach AI agents how to use this MCP server effectively. To install a skill, symlink it into your Claude skills directory:

ln -s /path/to/mcp-server/skills/clamm-liquidity ~/.claude/skills/clamm-liquidity

Available skills:

Reporting Issues

Found a bug, missing feature, or unclear documentation while using this server? Open an issue: https://github.com/arcadia-finance/mcp-server/issues

For AI agents: If you hit a gap during task execution, open an issue with:

• Tool name and parameters you used
• What you expected vs what happened
• The strategy or goal you were trying to execute

Pull requests for documentation fixes are welcome from agents and humans alike. All PRs require human review before merging.

Where to Find Us

• npm: @arcadia-finance/mcp-server
• MCP Registry: registry.modelcontextprotocol.io
• Smithery: smithery.ai
• MCP Servers: mcpservers.org
• LobeHub: lobehub.com
• PulseMCP: pulsemcp.com
• MCP Market: mcpmarket.com
• Glama: glama.ai
• awesome-mcp-servers: github.com/TensorBlock/awesome-mcp-servers

Development

yarn dev          # Run with tsx (hot reload)
yarn build        # Compile TypeScript
yarn test         # Run tests
yarn lint         # Lint with ESLint
yarn format       # Check formatting with Prettier