TrustGate MCP Server

A trust layer for AI trading agents. Before an agent buys a token, it calls TrustGate and gets a legitimacy score, a tier, and a plain trade verdict. Scams and low-trust tokens get flagged before any capital moves.

This is an MCP server exposing a single tool, trustgate_score. It runs on its own, alongside the Bitget Agent Hub MCP server or any other. Your agent sees both toolsets in one session: Bitget's trading tools and TrustGate's risk check.

Built for the Bitget AI Base Camp Hackathon S1, Trading Infra track.

What it does

TrustGate scores tokens from on-chain behavioral signals (holder distribution, deployer history, trading patterns, liquidity, bot activity). The score is a single number 0 to 100 with a tier on top of it.

| Tier | Score | Verdict | What the agent should do | | --- | --- | --- | --- | | HIGH_ELITE | 80 to 100 | ok | Trade. Strong trust signals. | | HIGH | 60 to 79 | ok | Trade. | | MEDIUM | 40 to 59 | caution | Trade only with a tight risk budget. | | LOW | 1 to 39 | avoid | Skip. High risk. | | BLOCKED | 0 | avoid | Skip. No usable history. | | NTT | n/a | not_tradeable | Not a token. NFT, plain contract, or wallet. |

The tool never exposes the internal scoring breakdown. Score and tier only, by design.

Install

No global install needed. The server runs straight from npm via npx.

You will register it as an MCP server in whatever agent you use. Pick your client below.

Claude Code

claude mcp add -s user trustgate -- npx -y trustgate-mcp-server

That sits next to your Bitget server. After adding both, the agent has Bitget's trading tools and trustgate_score in the same session.

Cursor

Add to your MCP config (.cursor/mcp.json or the global one):

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

Note: Cursor caps total tools around 40. The Bitget default load is 36, so adding this one tool is fine. Don't load Bitget's full module set at the same time or you'll go over.

Codex

Add to ~/.codex/config.toml:

[[mcp_servers]]
name = "trustgate"
command = "npx"
args = ["-y", "trustgate-mcp-server"]

The tool

trustgate_score

| Param | Type | Required | Description | | --- | --- | --- | --- | | address | string | yes | Token contract address to score. | | chain | string | no | Chain the token lives on. Defaults to arc. Informational. |

Returns a short text summary plus a structured result:

{
  "address": "0x...",
  "chain": "arc",
  "score": 84,
  "tier": "HIGH_ELITE",
  "verdict": "ok",
  "summary": "TrustGate score for 0x... on arc: 84/100, tier HIGH_ELITE. Looks legitimate. Safe to trade on trust grounds."
}

How it fits a trading agent

The point is the gate. A typical agent loop on Bitget:

1. A signal fires (a token is pumping, sentiment flips, an on-chain event hits).
2. Before routing the buy, the agent calls trustgate_score on the token.
3. If the verdict is avoid or not_tradeable, the agent skips and logs why.
4. If ok or caution, it sizes the position and trades through the Bitget tools.

One call between perception and execution. That's the whole integration. No change to how the agent trades, just a check it can't easily skip.

Configuration

| Env var | Default | Purpose | | --- | --- | --- | | TRUSTGATE_BASE_URL | https://trustgated.xyz | Point at a staging deploy if needed. | | TRUSTGATE_TIMEOUT_MS | 8000 | How long to wait on a score before treating the token as unverified. |

If the endpoint is slow or unreachable, the tool fails soft: it returns tier UNKNOWN and verdict unknown rather than throwing, so the agent can decide how to handle an unverified token instead of crashing the loop.

Build from source

pnpm install   # or npm install
pnpm build     # compiles src to dist
node dist/index.js

Requires Node 18 or newer.

License

MIT