close
Safety checks for Solana bots and AI agents

Check before your bot acts.

Rug-check scores and a live pump.fun launch feed, plus pre-sign, prompt-injection and x402 checks over MCP. One HTTP call, paid per request with x402. No account, no API key, no subscription.

Run a live check Read the docs
0.002USDC per rug-check
–launches in feed now
–flagged high risk
Burned from API payments
0$AFUEL

The counter starts at launch.
agentfuel · live demo
try: 
# Press Check to score a token.
# Your bot would call:
GET /v1/rugcheck/<mint>
402 Payment Required · 0.002 USDC
# pay from its own wallet, retry with PAYMENT-SIGNATURE
200 OK
– RISK 0–100
waiting
top 10 hold–
Demo runs on the free MCP trial (10 calls a day). Bots pay per call over x402.Not financial advice.
Live from Solana mainnet

Every new pump.fun launch, scored as it lands

This is the real launch feed, straight from the API. A sniper bot reads the same data and skips anything above its risk limit.

LaunchedTokenDev buyTop 10RiskMain finding
Loading the latest launches…

Refreshes every 30 seconds while this tab is open. Scores are automated on-chain checks; a low score does not mean a token is safe.

How it works

Pay per call, machine to machine

x402 puts the payment inside the HTTP request. Your bot needs a wallet, not a sign-up form. Click a step to follow the call.

Your bot AgentFuel Facilitator GET /v1/rugcheck/… 402 · price + pay-to retry + signed payment verify settle on success 200 · risk score
Pricing

Fractions of a cent, only for what you use

No plans and no minimum. Work out what your bot would spend.

1,000
1,440
Per month, about
$0USDC
Rug-checks
–
Launch feed
–
Per call
0.002 / 0.001
Integrate

Five minutes from copy to first score


        

      

    

  


  

    

      
How we work
Built to be checked

      

        
No custodyBots pay from their own wallet. We never hold anyone's funds.

        
Open sourceThe API and every scoring rule are on GitHub, so you can see why a token scored what it did.

        
Machine-readableThe manifest at /v1 lists every endpoint and price, so agents can discover the API themselves.

        
Public numbersChecks served today and in total, paid calls in USDC and $AFUEL, and the $AFUEL burn are shown live at the top of this page (/v1/stats).

      

    

  


  

    

      
Partners
Checks we run together

      

        FizzlMaker of Presign Guard and x402 Doctor. AgentFuel's MCP server passes those checks on to them. More agent tools at fizzl.eu.
        Presign GuardChecks a transaction before your bot signs it, and spots prompt injection. Powers safe_buy, presign_check and injection_check.
        Jev by TypeSafeReads a token's name, ticker and description, which on-chain checks cannot see. Flags tokens that pose as a known brand or person, copy a well-known coin, or use scam wording. Part of every rug-check and the launch feed.
        x402 DoctorChecks another x402 endpoint before your agent pays it. Powers x402_check.