RouteX402
# routex402-mcp
MCP server for **Routex402** — verified payment reputation for AI agents on Solana. Drop into Claude Desktop, Cursor, VS Code, or any MCP-compatible client, and you can ask the model to check a seller before paying it.
> **Status:** Solana **devnet** only. See [routex402.xyz](https://routex402.xyz).
## Install (Claude Desktop)
Add this to `~/Library/Application Support/Claude/claude_desktop_config.json` (macOS) or `%APPDATA%\Claude\claude_desktop_config.json` (Windows):
```json
{
"mcpServers": {
"routex402": {
"command": "npx",
"args": ["-y", "routex402-mcp"]
}
}
}
```
Restart Claude Desktop. The tools `check_agent_reputation` and `list_top_agents` appear in the tool picker.
## Try it
Ask Claude:
> "Should I pay seller `CiYKR8EYEJvJ6sGJNwYmKWSFdCBC3PsZSMYQao6aHk6J`? Use the strict policy."
Claude calls `check_agent_reputation` and returns the score, vetting checks, and a proceed / do_not_pay recommendation.
## Tools
### `check_agent_reputation(seller, policy?)`
Returns a full agent report with score, tier, confidence, five vetting checks (track record, success rate, buyer independence, volume consistency, recency), and a policy decision. Read-only.
### `list_top_agents(limit?)`
Returns sellers ranked by reputation, highest first.
## Environment
- `ROUTEX402_BASE_URL` — override the API endpoint (default: `https://route402-seller.onrender.com`).
## License
MIT
TDQS
Scored across 2 tools
The two tools have clearly distinct purposes: one performs a targeted lookup for a specific agent, while the other returns a ranked discovery list. An agent can easily choose between checking a known seller and browsing top sellers.
Both names follow a consistent snake_case verb_noun pattern: check_agent_reputation and list_top_agents. The verbs and object nouns are predictable and readable.
Two tools is thin for a reputation/discovery server, even if both are useful. It covers the two headline workflows but leaves little room for filtering, detail lookup, or supporting operations.
The core workflows of checking a known agent and listing top agents are covered, but the surface lacks search/filtering, detailed agent profiles or history, and any mechanism for submitting or updating reputation data. These are notable gaps for a reputation-oriented server.