MCP server · @solsentry/mcp v0.2.3

SolSentry in your AI agent.

Install once, query forever. Six tools. Works in Claude Code, Cursor, Windsurf, Zed, or anything that speaks the Model Context Protocol. Your AI now knows every known Solana rug operator and can cite them by wallet.

npm install →Setup guide Tool reference

Install

One command. Two config lines.

Claude Code / Claude Desktop

Add to ~/.claude/claude_desktop_config.json or your project’s .mcp.json:

{
  "mcpServers": {
    "solsentry": {
      "command": "npx",
      "args": ["-y", "@solsentry/mcp"]
    }
  }
}

Cursor

Add to ~/.cursor/mcp.json or workspace settings:

{
  "mcp.servers": {
    "solsentry": {
      "command": "npx",
      "args": ["-y", "@solsentry/mcp"]
    }
  }
}

First use

Ask your agent:

  "Use solsentry.check_token to check
   6n6iS9z5Lz3zMfrGHascQov4ZJu939DRVLUpmAjC55ch,
   then summarize the risk verdict."

The agent calls the tool, gets back JSON,
and writes you a human answer. No setup
beyond the MCP server install.

6 tools · all public

What the MCP exposes

Every tool hits the production API at api.solsentry.app. Data is live, not cached beyond 30 seconds. No API key required for public tools.

solsentry.scan_wallet

(wallet: string)

Lookup operator profile — known flag, risk level, confirmed rugs, total tokens, rug rate, tags.

solsentry.scan_token

(mint: string)

Token analysis — risk, flags, dev wallet, bot cluster links, outcome if resolved.

solsentry.recent_alerts

(limit?: number)

Latest HIGH and CRITICAL alerts with mint, dev, and flag list.

solsentry.recent_resolutions

(limit?: number)

Outcome stream — which predictions were validated and which missed.

solsentry.drain_trace

(wallet: string)

SOL flow trace, up to 10 hops, with bridge + CEX classifications.

solsentry.network_stats

(—)

Global counters — total scans, accuracy, runtime, confirmed rugs.

Why MCP matters

Your AI already has context. Give it intel.

Claude, ChatGPT, and Cursor now drive most dev workflows. MCP is how you plug live on-chain intelligence into them. SolSentry sits next to any general-purpose agent and answers the question it can't answer alone: who deployed this token, and what did they do last time?

For traders

Paste a wallet in Claude, ask 'is this a serial rugger?', get back confirmed_rugs + rug_rate + recent alerts. No tab switching.

For security researchers

Ask your agent to trace a drain, then summarize the cash-out path and flag any known CEX endpoints. 10 hops, one call.

For builders

Embed SolSentry intel into code-gen workflows. Your agent validates operator risk before suggesting a swap path or listing integration.

Troubleshooting

If the tool does not appear

Restart the agentMCP servers load at startup. Re-launch Claude Code / Cursor after editing the config.
Verify Node is availablenpx needs Node 18+. Run npx @solsentry/mcp --version in a terminal to confirm the binary installs.
Check the API is reachablecurl https://api.solsentry.app/health should return 200 OK. If not, the MCP will surface a network error on first tool call.
Still stuck?Reach out — hello@solsentry.app or @solsentryai. MCP is early — bug reports are welcome.