# AgentFuel

> Safety checks for Solana (and Robinhood Chain) bots and AI agents: "Check before your bot acts." Rug-check risk scores for any Solana token, a live feed of new pump.fun launches with risk scores, and pre-sign, prompt-injection and x402 endpoint checks. One HTTP API and one MCP server. Bots pay per call with x402 (USDC on Solana); no account, no API key, no subscription.

- API base: https://api.agentfuel.eu
- MCP server (Streamable HTTP): https://api.agentfuel.eu/mcp — tools: rugcheck, rugcheck_robinhood, latest_launches, safe_buy, presign_check, injection_check, x402_check, afuel_token. safe_buy gives one green/orange/red verdict before a buy (rug-check on the token plus a pre-sign check on the transaction). 50 free calls per IP per day.
- Prices: 0.002 USDC per rug-check, 0.001 USDC per launch-feed call, 0.01 USDC per safe-buy check. Always current in the manifest at https://api.agentfuel.eu/v1
- Official MCP registry name: io.github.Fizzl13/agentfuel

## Docs

- [Developer docs](https://www.agentfuel.eu/docs.html): quickstart, MCP setup, x402 payment flow, endpoints, responses, errors and limits
- [Quickstart](https://www.agentfuel.eu/docs.html#quickstart)
- [Use from an AI agent (MCP)](https://www.agentfuel.eu/docs.html#mcp)
- [Paying with x402](https://www.agentfuel.eu/docs.html#x402)
- [How the risk score works](https://www.agentfuel.eu/docs.html#scoring)
- [How we compute a rug score (blog)](https://www.agentfuel.eu/blog/rug-score.html): every rule and weight behind the 0 to 100 score
- [Check the token, then check the transaction (blog)](https://www.agentfuel.eu/blog/safe-buy.html): safe_buy, rug-check plus pre-sign check in one verdict
- [Give your AI agent a rug-check in two minutes (blog)](https://www.agentfuel.eu/blog/mcp-rugcheck.html): MCP setup and a prompt rule

## Machine-readable

- [Agent manifest](https://api.agentfuel.eu/v1): endpoints, prices, tiers (JSON, free)
- [OpenAPI document](https://api.agentfuel.eu/openapi.json)
- [x402 discovery](https://api.agentfuel.eu/.well-known/x402)

## Endpoints

- [GET /v1/rugcheck/{mint}](https://www.agentfuel.eu/docs.html#rugcheck): risk score 0-100 for a Solana token (mint/freeze authority, holder concentration, vesting locks, pump.fun Mayhem Mode).. A 0x address is checked on Robinhood Chain instead (same as /v1/rugcheck/robinhood/{token}); the MCP rugcheck tool works the same way. Paid with x402.
- [GET /v1/rugcheck/robinhood/{token}](https://www.agentfuel.eu/docs.html#robinhood): the same risk score for an ERC-20 token on Robinhood Chain (chain id 4663): verified source, proxy, owner, mint/blacklist/fee functions, holder concentration. Paid with x402, same price as the Solana rug-check.
- [GET /v1/launches](https://www.agentfuel.eu/docs.html#launches): newest pump.fun launches with creator, dev-buy share and risk score. Paid with x402.
- [POST /v1/safe-buy](https://www.agentfuel.eu/docs.html#safe-buy): one green/orange/red verdict before a buy. JSON body {mint, transaction (base64, unsigned), intent?, exclude?}; runs the rug-check plus Presign Guard's transaction check. 0.01 USDC with x402; a bad body gets 400 and is not charged.
- [GET /v1/token](https://api.agentfuel.eu/v1/token): the official $AFUEL mint address, launch wallet and links (pump.fun, Solscan). Free. Official mint: 8Qj1i5Q8DJ1HJW2rEpdzED47jNi2f5XvaqwykqQQpump (launched 2026-11-07, launch wallet GjxNqj5tf4rUaz3XHagxPoRqhCf7Y9WLJhNdKtBkSvkT). Any other mint is not AgentFuel. Also the MCP tool afuel_token.
- [GET /v1/token/burn](https://api.agentfuel.eu/v1/token/burn): burned $AFUEL and current supply, read on-chain. Free.
- [GET /v1/token/payments](https://api.agentfuel.eu/v1/token/payments): paid API calls and USDC received, read on-chain. Free.

## Code


## Optional

- [$AFUEL token facts](https://www.agentfuel.eu/token.html): access token for the API. Holding it gives free daily calls and higher rate limits. No yield. Can lose all value.
- [Litepaper](https://www.agentfuel.eu/litepaper.html)
- [More agent tools by Fizzl](https://fizzl.eu): Presign Guard, x402 Doctor
- [Jev by TypeSafe](https://typesafe.ai): AI name check inside every rug-check and the launch feed: flags impersonation of a brand or person, clones of well-known coins and scam wording (warn findings NAME_IMPERSONATION, NAME_CLONE, SCAM_WORDING at 80% or more).
- Rug-check scores are automated on-chain checks, not financial advice.