Mammoth MCP Server
Enables LLM agents to interact with the Mammoth Protocol on Solana for discovering cycles, getting quotes, and executing buys via bounded-curve mechanics.
README
@mammoth-protocol/mcp-server
MCP (Model Context Protocol) server for Mammoth Protocol.
Makes Mammoth a native tool for any LLM agent — Claude, GPT, Gemini, Eliza, Virtuals, ai16z, and any other MCP-compatible system. No custom integration per agent. One server, all of them.
Quick Start
npx @mammoth-protocol/mcp-server
Read-only by default (discovery, quotes, snapshots). Add a wallet for buy execution:
MAMMOTH_WALLET_KEY='[your-keypair-json-array]' npx @mammoth-protocol/mcp-server
Connect to Claude Desktop
Add to ~/Library/Application Support/Claude/claude_desktop_config.json:
{
"mcpServers": {
"mammoth": {
"command": "npx",
"args": ["@mammoth-protocol/mcp-server"],
"env": {
"SOLANA_CLUSTER": "devnet"
}
}
}
}
With buy execution:
{
"mcpServers": {
"mammoth": {
"command": "npx",
"args": ["@mammoth-protocol/mcp-server"],
"env": {
"SOLANA_CLUSTER": "mainnet-beta",
"SOLANA_RPC_URL": "https://your-rpc-endpoint.com",
"MAMMOTH_WALLET_KEY": "[your,keypair,as,json,array]"
}
}
}
}
Connect to Cursor / Windsurf / Cline
Add to your MCP config file (.cursor/mcp.json or equivalent):
{
"mcpServers": {
"mammoth": {
"command": "npx",
"args": ["@mammoth-protocol/mcp-server"]
}
}
}
Connect to a Custom Agent (Eliza, Virtuals, ai16z)
const { Client } = require('@modelcontextprotocol/sdk/client/index.js');
const { StdioClientTransport } = require('@modelcontextprotocol/sdk/client/stdio.js');
const transport = new StdioClientTransport({
command: 'npx',
args: ['@mammoth-protocol/mcp-server'],
env: { SOLANA_CLUSTER: 'devnet' },
});
const client = new Client({ name: 'my-agent', version: '1.0.0' });
await client.connect(transport);
// List all open cycles
const result = await client.callTool({ name: 'mammoth_get_open_cycles', arguments: {} });
console.log(result.content[0].text);
// Get snapshot for a specific project
const snap = await client.callTool({
name: 'mammoth_get_cycle_snapshot',
arguments: { mintAddress: 'YOUR_MINT_ADDRESS' }
});
// Get a buy quote
const quote = await client.callTool({
name: 'mammoth_buy_quote',
arguments: { mintAddress: 'YOUR_MINT_ADDRESS', solAmount: 0.5 }
});
Available Tools
| Tool | Description |
|---|---|
mammoth_get_open_cycles |
All active cycles with price, fill %, and curve type |
mammoth_get_cycle_snapshot |
Full decision snapshot for a specific mint |
mammoth_get_projects |
All Mammoth projects with supply mode and cycle count |
mammoth_get_rights_window_cycles |
Cycles where rights window is currently open |
mammoth_check_rights |
Check if a wallet has unexercised rights |
mammoth_buy_quote |
Calculate tokens out for SOL in — no transaction |
mammoth_buy_tokens |
Execute a buy (requires MAMMOTH_WALLET_KEY) |
mammoth_protocol_reference |
Full protocol mechanics and parameter reference |
Available Resources
| URI | Description |
|---|---|
mammoth://protocol/reference |
Protocol constants and parameter definitions |
Environment Variables
| Variable | Default | Description |
|---|---|---|
SOLANA_RPC_URL |
Devnet public RPC | RPC endpoint (use a private endpoint for production) |
SOLANA_CLUSTER |
devnet |
devnet or mainnet-beta |
MAMMOTH_WALLET_KEY |
none | Keypair as JSON array or base58 string. Required for buy execution. |
Why Mammoth for AI Agents
Mammoth Protocol is designed for autonomous participation:
- Bounded curves — deterministic pricing. A bot calculates exact cost before submitting. No AMM slippage surprises.
- Cycle events — agents only need to act at defined moments (cycle open, rights window, close). No constant monitoring required.
- Rights system — a bot holding from Cycle 1 gets guaranteed first access to Cycle 2 at base price. Structurally advantaged position.
- AuthorityConfig — a principal (human or AI) can delegate specific operations to an operator bot with explicit permissions and spending limits.
No other launchpad was built for this.
For Portfolio AI Agents
If you're building an AI that manages a user's high-growth crypto allocation:
- Call
mammoth_get_open_cyclesto discover available cycles - Call
mammoth_get_cycle_snapshoton interesting projects - Call
mammoth_buy_quoteto calculate exact outcome - Present to user or execute autonomously (if authorized)
- Subscribe to
CycleOpenedevents via the SDK for zero-latency triggers
The whole flow is programmatic. No UI. No human clicking buttons.
SDK
For direct programmatic access (without MCP):
npm install @mammoth-protocol/sdk
Links
- Protocol: https://mammoth-protocol.vercel.app
- AI Reference: https://mammoth-protocol.vercel.app/ai-reference
- Protocol Reference: https://mammoth-protocol.vercel.app/protocol
- SDK: https://github.com/kelvinsinferno/mammoth-sdk
MIT — Kelvinsinferno Studio
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。