pathguard

pathguard-mcp

PathGuard MCP lets MCP-compatible AI agents check crypto transactions for scams and mistakes directly in conversation. It supports local MCP clients such as Claude Desktop, Claude Code, Gemini CLI, Grok CLI, Cursor, and Codex, plus remote Streamable HTTP deployments.

Listed on the official MCP Registry as org.cieltech.pathguard/pathguard.

The production MCP server is a thin, authenticated interface to PathGuard's transaction-safety API. PathGuard analyzes transactions before a send; it does not sign, broadcast, or transfer funds.

Production MCP

Remote Streamable HTTP endpoint:

https://pathguard-v2.up.railway.app/mcp/

Remote clients use OAuth to connect an existing PathGuard account. During authorization, PathGuard verifies the user's existing API key. The API key is not exposed to the AI client or returned through MCP.

For local/stdio use, the API key is supplied to the local server process through PATHGUARD_API_KEY.

Install

The package is currently installed directly from this public repository and is not published to PyPI yet.

git clone https://github.com/Utee/pathguard-mcp.git
cd pathguard-mcp
python -m pip install .

For development, use an editable install:

python -m pip install -e .

Requires Python 3.10+ and mcp >= 2.0.

Get an API key

Create a PathGuard account and API key from the PathGuard API documentation.

Keep your API key private. Do not commit it to GitHub or expose it in browser-side code.

Local MCP clients

For local clients, the server uses stdio and authenticates to the PathGuard API with PATHGUARD_API_KEY.

Claude Desktop

Add this to your Claude Desktop config:

{
  "mcpServers": {
    "pathguard": {
      "command": "pathguard-mcp",
      "env": {
        "PATHGUARD_API_KEY": "pg_your_key_here"
      }
    }
  }
}

Claude Code

claude mcp add --transport stdio pathguard -- pathguard-mcp

Set PATHGUARD_API_KEY in the environment that starts Claude Code.

Gemini CLI

{
  "mcpServers": {
    "pathguard": {
      "command": "pathguard-mcp",
      "env": {
        "PATHGUARD_API_KEY": "pg_your_key_here"
      }
    }
  }
}

Grok CLI

export PATHGUARD_API_KEY=pg_your_key_here
grok mcp add pathguard -- pathguard-mcp

Codex, Cursor, VS Code, and other local MCP clients

Use the same stdio server definition as the Claude Desktop example. The exact configuration location varies by client.

Remote clients

For remote clients, use the production OAuth-protected Streamable HTTP endpoint shown above. Never put a PathGuard API key in browser-side configuration.

ChatGPT

In a ChatGPT workspace with Developer Mode enabled, create a custom MCP app and provide:

https://pathguard-v2.up.railway.app/mcp/

Choose OAuth authentication. ChatGPT will use the PathGuard authorization flow to connect an existing PathGuard account.

The authorization flow verifies the user's PathGuard API key, then issues OAuth credentials for the MCP connection. The raw PathGuard API key is not returned to ChatGPT.

Available tools

Tool	What it does
check_transaction	Inspect one crypto destination and optional amount for PathGuard risk signals.
preflight_transaction	Run the full pre-send check and return ALLOW, REVIEW, or BLOCK.
check_transactions_batch	Check up to 1,000 transactions and return a compact risk summary or full results.
report_scam_address	Record a suspected scam address when the user explicitly asks to report it.
get_usage_status	View the connected account's plan and monthly scan usage.
get_profile	Identify the PathGuard account represented by the OAuth connection.

Tool safety boundary

PathGuard MCP is analysis-only. Its tools do not:

• sign transactions;
• broadcast transactions;
• transfer funds;
• request private keys or seed phrases;
• replace your application's authorization controls.

A BLOCK result is a PathGuard safety signal. The wallet, application, or agent remains responsible for authorization, signing, and final transaction execution.

Batch checking

check_transactions_batch supports up to 1,000 items.

By default it returns a compact summary containing:

• total items;
• processed items;
• ALLOW / REVIEW / BLOCK counts;
• maximum risk score;
• flag counts;
• a sample of flagged transactions.

Use detail_level="full" when individual results are required.

Security

• Keep API keys server-side and private.
• Use HTTPS for remote deployments.
• OAuth protects the production remote MCP connection.
• The production MCP server maps the authenticated OAuth subject to an existing PathGuard account.
• Internal server-to-server authentication is not exposed as an MCP argument or model-visible credential.
• Never commit API keys, production credentials, or local .env files.

Full API reference

See the PathGuard API documentation for endpoint schemas, rate limits, pricing, and risk-flag information.

PathGuard MCP: https://pathguard.cieltech.org/mcp