Overview
bitget-agent-mcp is the official Model Context Protocol (MCP) server for Bitget, enabling desktop AI agents like Claude Desktop, Cursor, , , and to operate your Bitget account through natural-language commands.
Power your agent
คำสั่งของผู้เผยแพร่เอง ไม่ใช่เอนด์พอยต์ของ PlugMCP
Host a copy with our keys and logs.
npx -y @bitget-ai/[email protected]Claude Code ไม่มีลิงก์ติดตั้งสำหรับคำสั่งแบบโลคัล ให้คัดลอกคำสั่งด้านล่างแล้วรันในเทอร์มินัล
claude mcp add bitget-ai-bitget-agent-mcp -- npx -y @bitget-ai/bitget-agent-mcp@3.3.1รันในเทอร์มินัล
{ "mcpServers": { "bitget-ai-bitget-agent-mcp": { "command": "npx", "args": [ "-y", "@bitget-ai/[email protected]" ] } }}claude_desktop_config.json
URL แบบรีโมตหรือคำสั่งบนเครื่องของคุณ? การติดตั้งสองแบบต่างกันอย่างไร
รันบน PlugMCP
A new server you create. It is not this listing's official endpoint. Use the publisher install when you want their URL and their credentials.
เปิดจากผู้เผยแพร่
bitget-agent-mcp is the official Model Context Protocol (MCP) server for Bitget, enabling desktop AI agents like Claude Desktop, Cursor, , , and to operate your Bitget account through natural-language commands.
Official Bitget MCP Server for the Unified Trading Account (UTA / v3) API — exposes a curated, progressively-discoverable intent surface (account, trade, market, strategy, …) to any MCP-compatible AI host (Claude Desktop, Cursor, Continue, …) over stdio.
It is built on the Bitget Unified Trading Account (UTA / v3) API and covers 89 trading operations across market data, spot, futures, account & funds management, sub-accounts, loans, and tax. Crucially, it does not flood your model with one tool per endpoint. Instead it exposes a small, progressively-discoverable intent surface of 14 curated intent verbs (the default profile loads 12 verbs + discover + raw = 14 tools) so AI hosts get full capability without the context bloat and tool-selection errors that plague endpoint-per-tool servers.
Part of Bitget Agent Hub — the official open-source AI ecosystem for Bitget, including the CLI, SDK, installer, and market-analysis skills.
Paste this prompt into your AI agent (Claude Desktop / Cursor / etc.):
Please configure the Bitget MCP Server for my AI tool (requires Node.js 20+):
first ask me which AI tool I use (Claude Desktop / Cursor / Windsurf / ChatGPT
Desktop), then add a server that runs `npx -y @bitget-ai/bitget-agent-mcp` to
that tool's MCP config, with BITGET_API_KEY, BITGET_SECRET_KEY, and
BITGET_PASSPHRASE filled in. Confirm once the Bitget tools have loaded.
The agent will detect your tool, add the npx server entry, wire up credentials, and verify the connection.
Prefer to edit config files yourself? Jump to Configuration for per-tool JSON snippets. The command is always:
npx -y @bitget-ai/bitget-agent-mcp
Once configured, your AI gains native access to Bitget's trading stack through natural language. A few examples (every one maps to a real operation in the surface):
Market data (the market verb) is public and works without API credentials; everything that touches your account requires keys.
--read-only — Blocks every write operation for the session. The verbs stay visible, but any order, transfer, cancellation, or withdrawal is rejected before it reaches Bitget. Ideal for safe exploration. (Mutually exclusive with --paper-trading.)--paper-trading — Routes signed requests to Bitget's Demo Trading environment (adds the paptrading: 1 header). Requires a separate Demo API Key. Perfect for rehearsing strategies risk-free.--modules <list> — Load only the modules you need. Default: account,trade,market. On-demand: strategy, cryptoloans, tax.--surface <intent|full> — intent (default) exposes the curated verbs. full additionally emits one tool per underlying v3 endpoint.The intent surface is progressively discoverable. Rather than reading every schema up front, the agent follows discover → drill down → execute:
discover({}) → list business domains + meta-tools
discover({ domain: "trade" }) → that domain's verbs, one line each
discover({ tool: "order" }) → one verb's full input schema (+ its actions)
discover({ tool: "order", action: "place" }) → one action's exact required/optional contract
order({ action: "place", ... }) → execute
If the prompt already implies the verb and arguments, the agent skips discovery and calls directly. discover({ search: "funding" }) keyword-searches the whole surface when the domain is unknown. The two meta-tools are always present:
discover — progressive introspection of the surface.raw — an escape hatch that reaches any v3 operation by operationId for the long tail. (In practice every operation is also covered by a verb, so raw is rarely needed.)cancelAll, withdraw) return { confirmationRequired: true } unless you pass confirm: true.dryRun: true to preview the would-send request without sending it.MCP tool annotations are derived from the SDK's riskLevel (read / write / high), so hosts can flag destructive operations before the model invokes them.
graph TD
A[AI Host<br/>Claude Desktop / Cursor / …] -->|MCP over stdio| B[bitget-agent-mcp<br/>thin protocol adapter]
B --> S[@bitget-ai/bitget-agent-sdk<br/>intent verbs · discover · raw<br/>write-safety gate · HMAC signing · retry]
S -->|HMAC-SHA256 signed| C[Bitget UTA v3 REST API<br/>api.bitget.com]
D[Environment Variables<br/>BITGET_API_KEY etc.] --> B
style B fill:#f9f,stroke:#333,stroke-width:2px
style C fill:#bbf,stroke:#333,stroke-width:2px
How it works:
npx over stdio — no network listener, no proxy.All the "smarts" — discovery, intent routing, the write-safety gate, signing, and response shaping — live in @bitget-ai/bitget-agent-sdk. This package is a thin stdio adapter on top of it.
API Key, Secret Key, Passphrase⚠️ Security note: store these securely. Never share them or commit them to version control.
Use the one-step prompt, or add the server to your host's MCP config manually (see Configuration).
Ask your AI: "What Bitget tools are available?" You should see verbs like market, order, position, account_overview, plus discover and raw.
Edit claude_desktop_config.json:
~/Library/Application Support/Claude/claude_desktop_config.json%APPDATA%\Claude\claude_desktop_config.json~/.config/Claude/claude_desktop_config.json{
"mcpServers": {
"bitget": {
"command": "npx",
"args": ["-y", "@bitget-ai/bitget-agent-mcp"],
"env": {
"BITGET_API_KEY": "your-api-key-here",
"BITGET_SECRET_KEY": "your-secret-key-here",
"BITGET_PASSPHRASE": "your-passphrase-here"
}
}
}
}
Restart Claude Desktop after saving.
Settings → MCP → Add New Server:
⚠️ Cursor tool limit: Cursor caps total MCP tools at 40 across all servers. The default Bitget profile loads 14 tools (12 intent verbs +
discover+raw), leaving 26 slots for other servers.--modules allyields 16 tools.
Use the same npx -y @bitget-ai/bitget-agent-mcp command and pass credentials via environment variables, following your tool's MCP configuration docs. If the host speaks MCP over stdio, it works.
Read-only mode (safe for exploration):
{ "args": ["-y", "@bitget-ai/bitget-agent-mcp", "--read-only"] }
Paper trading (Demo environment — use Demo API keys):
{
"args": ["-y", "@bitget-ai/bitget-agent-mcp", "--paper-trading"],
"env": {
"BITGET_API_KEY": "your-demo-api-key",
"BITGET_SECRET_KEY": "your-demo-secret-key",
"BITGET_PASSPHRASE": "your-demo-passphrase"
}
}
Load extra modules:
{ "args": ["-y", "@bitget-ai/bitget-agent-mcp", "--modules", "account,trade,market,cryptoloans,tax"] }
bitget-agent-mcp [options]
--modules <list> account, trade, market, strategy,
cryptoloans, tax
"all" loads every module.
Default: account,trade,market
--surface <mode> intent curated verbs + discover + raw (default)
full ALSO emit one tool per underlying v3 endpoint
--read-only Expose only read/query operations; block all writes.
--paper-trading Enable Demo Trading mode (requires a Demo API Key).
Mutually exclusive with --read-only.
--help Show help and exit
--version Show version and exit
Without API credentials, only public/read (market data) operations succeed.
The default profile loads account,trade,market. The full set spans 6 modules and 14 curated intent verbs:
Always present regardless of module: discover (introspection) and raw (reach any v3 operation by operationId).
--read-only — blocks all write operations for the session; the AI can query but cannot place orders, transfer funds, cancel, or withdraw.--paper-trading — routes signed requests to Bitget's Demo environment. No real funds involved. Requires a separate Demo API Key.confirm: true, and any write supports dryRun: true for a no-network preview.The SDK applies a client-side retry/rate-limit policy that protects against AI loops hammering Bitget's API. If Bitget returns a 429, it backs off and retries automatically.
Cause: Node.js not installed or not on PATH. Fix: install Node.js ≥ 20 from nodejs.org and restart your terminal/host.
Likely causes & fixes:
--paper-trading (or vice versa) → use a Demo API Key for paper trading.Cause: MCP server not configured correctly, or the host wasn't restarted. Fix: validate the config file is valid JSON, restart the host completely, then ask "List available MCP tools."
Cause: other MCP servers consuming Cursor's 40-tool cap. The Bitget default profile is 14 tools.
Fix: remove unused MCP servers, or use --modules to load only what you need.
Cause: too many rapid calls. Fix: the SDK auto-throttles, but if you hit Bitget's server-side limit, wait ~60s before retrying. Avoid asking the AI to place hundreds of orders in quick succession.
Check: Node.js version (node --version ≥ 20), network reachability to api.bitget.com, and that outbound HTTPS isn't blocked by a firewall.
Because installation uses npx -y, the latest version is pulled automatically each time your AI host starts — no manual updates needed.
To force-refresh the npm cache:
npx @bitget-ai/bitget-agent-mcp@latest --version
To upgrade the whole Bitget AI toolkit at once, use the installer in agent_hub.
It's the official MCP (Model Context Protocol) server for Bitget. It exposes 89 Bitget UTA v3 operations to desktop AI clients (Claude Desktop, Cursor, Windsurf, ChatGPT Desktop, …) through 14 curated intent verbs plus a discover introspection tool and a raw escape hatch — over the MCP standard.
bitget-agent-mcp — for desktop AI hosts that speak MCP (Claude Desktop, Cursor, …).agent-cli (bgc) — for terminal AI (Claude Code, Codex CLI) that runs commands in your shell.Both sit on the same SDK and expose the same intent surface — pick the one that matches your AI tool.
Endpoint-per-tool servers advertise one tool per endpoint, which bloats the model's context and degrades tool-selection accuracy. The intent surface keeps the default tool list at 14 tools (12 verbs + discover + raw) while still covering all 89 default-exposed operations directly — and raw reaches any operation by operationId when you need the long tail. The agent uses discover to drill into detail only when it needs to.
Default: account (39 ops), trade (17 ops), market (16 ops) = 72 operations, exposed through 12 intent verbs plus discover and raw (14 tools total). Load more with --modules strategy,cryptoloans,tax or --modules all (which reaches the 14 verbs across 89 endpoints).
Yes — Cursor caps total MCP tools at 40. The default Bitget profile loads 14 tools, well within the cap (26 slots free). --modules all yields 16 tools.
Add "--read-only" to your config args. Every write operation is rejected for the session — the AI can query balances and markets but cannot place orders, transfer funds, cancel, or withdraw.
Yes — use paper trading: create a Demo API Key, set the Demo credentials as environment variables, and add "--paper-trading" to your args. Signed requests route to Bitget's Demo environment.
Yes. Credentials live only in your host's MCP config, are never proxied through any server, are signed locally with HMAC-SHA256 before reaching Bitget, and are never logged or written to disk by this server.
Yes — MIT-licensed and free for personal and commercial use. The MCP server is maintained by Bitget at no cost.
Claude Desktop, Cursor, Continue, ChatGPT Desktop, Windsurf, and any MCP client that communicates over stdio transport.
Issues and pull requests are welcome.
MIT License — free for personal and commercial use.
⚠️ Risk disclaimer: trading cryptocurrency carries substantial risk. You are solely responsible for any orders your AI agent places on your behalf. Use --read-only and --paper-trading to rehearse safely before going live. Past performance does not guarantee future results.
Official Bitget Agent Hub tool · Part of the Bitget open-source AI ecosystem · Foundation: agent-sdk · Other surfaces: agent-cli · agent-skill · Market signals: bitget-signal
Official Bitget MCP Server for the Unified Trading Account (UTA / v3) API — exposes a curated, progressively-discoverable intent surface (account, trade, market, strategy, …) to any MCP-compatible AI
{ "mcpServers": { "bitget-ai-bitget-agent-mcp": { "command": "npx", "args": [ "-y", "@bitget-ai/[email protected]" ] } }}ภาพรวม
Official Bitget MCP Server for the Unified Trading Account (UTA / v3) API — exposes a curated, progressively-discoverable intent surface (account, trade, market, strategy, …) to any MCP-compatible AI host (Claude Desktop, Cursor, Continue, …) over stdio.
เครื่องมือ
ยังไม่มีรายการเครื่องมือ เซิร์ฟเวอร์แบบรีโมตจะได้รายการหลังการตรวจสอบสถานะครั้งแรก
FAQ
คัดลอกคอนฟิก stdio สำหรับ Cursor, Claude Desktop, Claude Code หรือ VS Code ได้ในหน้านี้ คอนฟิกนี้ใช้การติดตั้งของผู้เผยแพร่ ไม่ใช่พร็อกซีของ PlugMCP
ระเบียนนี้ไม่ได้ระบุตัวแปรสภาพแวดล้อมที่จำเป็น
@bitget Ai/bitget Agent ทำงานแบบโลคัลผ่าน stdio
ตรวจสอบล่าสุดเมื่อ 2026-11-18 Official Bitget MCP Server for the Unified Trading Account (UTA / v3) API — exposes a curated, progressively-discoverable intent surface (account, trade, market, strategy, …) to any MCP-compatible AI host (Claude Desktop, Cursor, Continue, …) over stdio.