Curvecall
Paid token market data, risk signals and trade simulations for AI agents. USDC via x402.
Pay-per-call market math and token intel for AI agents. Quote a trade, check slippage and get rug-risk signals in one paid call. No API keys, no signup. Agents pay $0.001-$0.02 per call via x402 v2: USDC on Base, plus USDC on Solana or shielded ZEC over the HTTP API. Built by arbonomous.
Live service · OpenAPI · MCP endpoint · llms.txt
Repo map: this public repo is the docs and discovery page for CurveCall. The service source code lives in a separate private repository and is not published here. Nothing in this repo runs the service.
CoinGecko tells you the price. CurveCall tells you what your buy will do.
How it works
sequenceDiagram participant A as Agent (MCP or HTTP client) participant C as CurveCall participant F as x402 facilitator A->>C: call a tool with no payment C-->>A: 402 + payment request A->>F: sign USDC payment (exact scheme) A->>C: retry with PAYMENT-SIGNATURE C->>F: verify C-->>A: 2xx data C->>F: settle only after a 2xx
Use it
- MCP (streamable HTTP):
https://curvecall.onrender.com/mcp/- 10 tools, same x402 gating - HTTP API:
https://curvecall.onrender.com- service card at/, OpenAPI at/openapi.json - Agent card (A2A):
https://curvecall.onrender.com/.well-known/agent.json - llms.txt:
https://curvecall.onrender.com/llms.txt
Call any endpoint without PAYMENT-SIGNATURE to get the 402 payment request (x402 v2, PAYMENT-REQUIRED header).
Tools
| Tool | Endpoint | Price |
|---|---|---|
quote_buy |
GET /v1/quote/buy?quote_reserves=&token_reserves=&amount_in= |
$0.001 |
quote_sell |
GET /v1/quote/sell?quote_reserves=&token_reserves=&amount_in= |
$0.001 |
token_snapshot |
GET /v1/token/{chain}/{address} |
$0.005 |
token_risk |
GET /v1/token/{chain}/{address}/risk |
$0.01 |
precheck |
GET /v1/precheck/{chain}/{address}?side=&amount_in= |
$0.01 |
contract_check |
GET /v1/token/{chain}/{address}/contract |
$0.01 |
wallet_snapshot |
GET /v1/wallet/{chain}/{address} |
$0.005 |
search_tokens |
GET /v1/search?q=&chain= |
$0.005 |
deployer_history |
GET /v1/deployer/{chain}/{address} |
$0.01 |
bundle_check |
GET /v1/token/{chain}/{address}/bundles |
$0.02 |
precheck is the one nobody else sells: full trade simulation (quote, price impact, slippage) plus rug-risk in a single call. contract_check reads the chain directly: ownership renounced, LP-token burn share, ERC-20 metadata. deployer_history answers "has this deployer rugged before?" from CurveCall's hourly launch-outcome index, with dev-behavior flags. bundle_check shows who bought in the first blocks after a launch, how much of the early supply they took, and whether any were funded by the deployer.
Chains: market data covers every chain Dexscreener indexes. On-chain reads cover base, ethereum, arbitrum, optimism, polygon and bsc. The deployer index and bundle/sniper signals cover Base, Robinhood Chain and Arc (bundles also BSC). All outputs are data and signals, never buy recommendations.
Plug it into your agent
Hermes Agent
Add to your Hermes MCP config (~/.hermes/config.yaml):
mcp_servers:
curvecall:
url: "https://curvecall.onrender.com/mcp/"
enabled: true
timeout: 120
OpenClaw
openclaw mcp set curvecall --url https://curvecall.onrender.com/mcp/ --transport streamable-http
or in the gateway config (json5):
{
mcp: {
servers: {
curvecall: {
url: "https://curvecall.onrender.com/mcp",
transport: "streamable-http",
},
},
},
}
Claude Desktop, Cursor, and other MCP clients
From the project's README.