100pro Token Risk Screen — MCP server
[100pro-token-risk MCP connector]
Pre-trade risk screening for EVM token contracts and Solana SPL mints, exposed as a remote MCP
tool. Honeypot/tax/owner flags, LP-lock state, mint & freeze authority, transfer fee/hook, holder
concentration, liquidity depth, wash-trade signals — ending in a one-line Verdict.
- Endpoint:
https://x402.rendraputra.dev/mcp (MCP streamable HTTP, JSON-RPC 2.0)
- Tool:
token_risk_screen (token, chain) — read-only, up to 5 comma-separated addresses per call
- Chains:
base, ethereum, bsc, polygon, arbitrum, solana
- Price: $0.10 USDC per single token, $0.25 for a 2-5 token batch, on Base, paid with
x402 v2 — no account,
no API key, no subscription. An unpaid call answers HTTP 402 with the payment challenge; pay it and
retry. Free metadata:
/llms.txt,
/openapi.json,
/.well-known/x402.
Use it as a remote MCP server (no install)
Use it through this stdio bridge
Some clients only speak stdio. This tiny proxy forwards initialize, tools/list and tools/call
to the remote endpoint (stdlib only, no dependencies):
or straight from the repo, if you use uv:
What a call returns
A markdown report per token, e.g.:
If the call is unpaid the server answers with the x402 payment challenge (HTTP 402). Pay $0.10 USDC
(single token) or $0.25 (2-5 token batch) on
Base to the payTo in the challenge and retry; the response carries a PAYMENT-RESPONSE receipt
naming the settled transaction.
Why paying beats a free scanner
Free scanners are the product being sold elsewhere: rate-limited, delay-baiting, or an upsell funnel.
This one is a single priced call with a fail-closed guarantee — the report is generated and returned
only after the USDC transfer is confirmed on Base, and every settle attempt is logged with the
on-chain verdict.
Links
MIT licensed.
Repo layout
100pro_mcp.py — repo-root entry point (python3 100pro_mcp.py, and the file container
checkers execute after cloning); it bootstraps pro100/bridge.py.
pro100/bridge.py — the actual stdio bridge: answers initialize / tools/list locally
(so introspection needs no network) and forwards tools/call to the hosted endpoint.
pyproject.toml — packages pro100, console script 100pro-mcp (so uvx --from git+https://github.com/rndrpp/100pro-mcp 100pro-mcp works).
Python 3.9+, standard library only — nothing to install for the plain-script path.