# Agent402.Tools > Pay-per-call web tools for AI agents, payable over **x402 or MPP** - the applied layer of Agentic Finance: agents that pay and get paid on their own (explainer: /agentic-finance). **First job: search the web and answer questions** (`/api/search`, `/api/answer`, `/api/search-news`) - then the long catalog of 500+ tools via `/api/find`: deterministic utilities, a metered model gateway on four wires (OpenAI chat, OpenAI Responses, Anthropic Messages and Gemini; for example `POST /v1/metered/chat/completions`, `POST /v1/metered/messages`) and finished report products. Call an endpoint, receive an HTTP 402 carrying both offers (x402 PAYMENT-REQUIRED and MPP WWW-Authenticate: Payment), and settle from your own wallet - USDC via x402, or MPP on Base/Celo (USDC) or Tempo (USDC.e or PathUSD, native), or by **card** on premium tools >= $0.50 (a card payment token over MPP's `stripe` charge method - no wallet, no stablecoin) - or, on 170 of the 625 tools, pay with proof-of-work (CPU) and skip the wallet entirely. No human, no signup, no API key: the payment is the identity. Per-call prices from $0.001 (the metered gateway quotes each request from its body) - most tools $0.001 to $0.02, with premium AI and media tiers higher and multi-tool packs $0.003 to $0.101; every price is in /api/pricing and quoted in the 402. Base URL: https://agent402.tools The full text, with every skill pack and every tool by category, is at https://agent402.tools/llms-full.txt. Terms are defined once at https://agent402.tools/glossary, and the category is explained at https://agent402.tools/agentic-finance. The complete machine-readable catalog with prices is https://agent402.tools/api/pricing. `tempo` charge (USDC.e or PathUSD on Tempo, settled natively via Tempo's own relay; not offered on identity-bound or long-running routes), then `evm` charge (USDC on Base or Celo, EIP-3009, settles on-chain identically to x402), then `stripe` charge (card, on routes priced $0.50 or more) **MPP: start here.** 20 fast, low-priced routes that offer the tempo challenge (USDC.e on Tempo) as well as the evm one: a good first call for a new MPP client. Prices are the catalog's; the full list is marked `x-mpp-flagship` in /openapi.json. - `GET /api/search-lite` ($0.008): Web search: up to five ranked results with title, URL and snippet. - `POST /api/extract` ($0.010): Read one known URL as clean markdown, with title, byline and word count. - `GET /api/crypto-price` ($0.001): Spot price, 24h change, volume and market cap for several coins in one call. - `POST /api/sol-price` ($0.001): USD price for up to 50 Solana token mints, with liquidity and the block read. - `POST /api/perp-funding` ($0.002): Current perpetual funding rate, annualized, plus recent hourly prints. - `POST /api/defi-yields` ($0.002): DeFi yield pools filtered by chain, project, token and TVL, sorted by APY or TVL. - `POST /api/kalshi-markets` ($0.002): Open prediction markets with yes/no prices, volume and resting size at the best price. - `GET /api/fx-rate` ($0.002): Convert an amount between two currencies at the reference rate, with its date. - `GET /api/treasury-yield-curve` ($0.005): The latest US Treasury yield curve, eleven tenors in percent. - `GET /api/unemployment-rate` ($0.005): Latest US unemployment rate plus a trailing monthly series. - `GET /api/edgar-company-lookup` ($0.001): Resolve a stock ticker to its SEC CIK and registered name, the key to every EDGAR call. - `GET /api/dns` ($0.001): DNS records for a domain: A, AAAA, MX, TXT, NS or CNAME. - `POST /api/asn-info` ($0.001): ASN, prefix, country and registry for an IP address or hostname. - `POST /api/tls-cert` ($0.001): A host's TLS certificate: issuer, validity window, days left, SANs. - `GET /api/weather-current` ($0.002): Current conditions anywhere on Earth, metric or imperial. - `GET /api/block-number` ($0.001): Latest block number on Ethereum, Base, Polygon, Arbitrum or Optimism. - `POST /v1/judge` ($0.001): A typed judgment about any content (choice, score or yes/no) with probabilities; model-backed. - `POST /api/time-convert` ($0.001): Convert between epoch, ISO 8601 and any IANA timezone. - `POST /api/hash` ($0.001): SHA-256, SHA-512, SHA-1 or MD5 of a text, in hex and base64. - `GET /api/uuid` ($0.001): Fresh UUIDs (v4 or v7): the simplest paid call to test an MPP client. **Batched option, no gas per call (Circle Gateway nanopayments).** On Base and Arc the same 402 also carries a second `exact` accept whose `extra.name` is `GatewayWalletBatched` (with `extra.verifyingContract` and a `maxTimeoutSeconds` of 604900 seconds): after a USDC deposit into Circle Gateway (an on-chain transfer the buyer makes and pays gas for), the buyer for each call signs the same EIP-3009 authorization under that domain against that balance and pays no gas for the call; Gateway verifies it and settles it off chain in batches. Same price, same payTo; a client that ignores `extra.name` keeps paying the plain accept, which stays first in `accepts`; agent402-client 0.8.20+ pays the batched one when constructed with `{ gateway: { signer } }`. Identity-bound routes and routes with a lasting effect (such as memory writes, attest and the route-execute tiers) offer the plain accept only. Machine-readable: `/api/pricing` (`payment.acceptKinds`, `payment.gatewayNanopayments`) and `/.well-known/x402`. ## Call a model (metered, the default) The model gateway's default route is metered: `POST /v1/metered/chat/completions`, `POST /v1/metered/messages`, `POST /v1/metered/responses` (OpenAI chat, Anthropic Messages and OpenAI Responses wires), no key and no account, from $0.001 a call. Model-backed: third-party models answer. The 402 quotes this exact request from its own body (the input estimated from its tokens with headroom, plus your `max_tokens`, at the model's list price plus a fee). On x402, Tempo and credits, settlement never exceeds that quote: an `exact` payer pays the quote, and an `upto` payer (where the 402 offers it) or a credits key settles what the answer used; a card payment adds the processor fee. ```bash curl -i -X POST https://agent402.tools/v1/metered/chat/completions -H 'content-type: application/json' \ -d '{"model":"anthropic/claude-haiku-4.5","messages":[{"role":"user","content":"Summarize x402 in one line"}],"max_tokens":200}' # unpaid: HTTP 402 quoting this body; an x402 or MPP client pays the quote and resends ``` Flat tiers stay as the predictable-price option: one price per call whatever the length, $0.003 to $0.50 (`/v1/nano`, `/v1`, `/v1/pro`, `/v1/premium`; `GET /v1/models` lists every model with its tier). ## Key machine surfaces - [/SKILL.md](https://agent402.tools/SKILL.md): agent-onboarding skill sheet - setup (MCP server / SDK / plain HTTP), discover, pay (x402, MPP or proof-of-work), read the 402, common issues. Start here if you are setting Agent402 up for the first time - [/api/search](https://agent402.tools/api/search): **front door** - live web search (title, URL, snippet). Start here to discover pages; follow with extract or answer - [/api/answer](https://agent402.tools/api/answer): **front door** - cited answer grounded in live web search results - [/api/search-news](https://agent402.tools/api/search-news): live news search for current events / headlines - [/api/find](https://agent402.tools/api/find): resolve a plain-language task to the best-matching tools with route, price, input schema, and a ready example (GET `?q={task}` or POST `{"task":"..."}`) - long-tail discovery behind the flagships. A task that names several steps ("get the price, then compute RSI") also returns a free `plan`: each step ranked the same way, with an estimated total - [/api/route](https://agent402.tools/api/route): Smart Order Router - rank tools across every x402 seller crawled from public registries; `include:"external"` excludes Agent402 for neutral cross-seller discovery - [/api/route/execute](https://agent402.tools/api/route/execute): the SOR that also PAYS. Send a task, and Agent402 resolves the best-matching tool, buys the result from the seller over x402 (any proven seller in the open index, not just ours), and sells it to you with a receipt - one payment, one request, one wallet. You never hold a wallet on their chain or sign up with them. `{"task":"...","include":"external"}`. Proportional tiers: $0.01 covers tools <= $0.005, `/api/route/execute-plus` at $0.05 covers <= $0.04, `/api/route/execute-max` at $0.55 covers <= $0.50, `/api/route/execute-pro` at $3.30 covers <= $3.00 - an over-cap task gets a self-correcting 409 naming the tier that fits - [/api/decide](https://agent402.tools/api/decide): a paid decision. Describe a job; get a call-ready plan over this catalog and outside x402 sellers with a recently verified 402: steps, fallbacks, params that validate against each tool's schema, cost and latency estimates. Depth quick $0.003 / plan $0.01 / full $0.03. The ranking formula has no first-party term and every tool carries firstParty. The fee returns as a 24-hour credit toward `POST /api/decide/execute`, which runs the plan when paid on Base or by credits or card (third-party steps at the seller's price plus a 5% markup). A decisionId is kept 30 days; execute it again with `"replan":true` to plan the saved task again on current prices and reliability first (the decision price for its depth is added to the quote; the answer's `replanned` field lists each changed step). Report outcomes free at `POST /api/decide/feedback`. - [/api/index](https://agent402.tools/api/index): the seller index, PAGINATED - one page, 250 max, never the whole set. The response says so: complete false, a Link header with rel=next, and sellerCount for the total. For ONE origin use ?seller=, which pages nothing and returns its full row with crawl history - [/api/leaderboard](https://agent402.tools/api/leaderboard): public on-chain ranking of x402 sellers by Base USDC settled volume, served as the TOP N and never the whole board - 25 rows by default, 50 the ceiling, and `totalSellers` carries how many are ranked in all, so a seller you cannot see in the rows may simply rank below them. Only sellers that SETTLED inside `windowServed` rank at all (pipeline: public x402 discovery registry → `eth_getLogs` on Base USDC → per-call ceiling filter, reported as `maxCallUsd` → aggregate by payTo; params `?sort=usd|calls`, `?top=N`, `?include=external|all`) - same data as the MCP tool `sellers.list` (listed when a wallet or credits key is set) and the `agent402-client` SDK method `topSellers()` - [/api/mpp-index](https://agent402.tools/api/mpp-index): the MPP seller index (live-verified WWW-Authenticate: Payment sellers with the payment offers their real 402 makes: method, recipient, currency, chain) - [/api/mpp-leaderboard](https://agent402.tools/api/mpp-leaderboard): on-chain ranking of MPP sellers by inbound USDC.e transfers on Tempo to their live recipient (window, distinct payers, volume; `routable` = the router will pay them) - [/.well-known/x402](https://agent402.tools/.well-known/x402): one-fetch service manifest (identity, payment options, capability map, MCP, trust signals) - [/.well-known/agent-card.json](https://agent402.tools/.well-known/agent-card.json): our A2A AgentCard (also at /.well-known/agent.json) - name, skills, transport. It declares HTTP+JSON rather than the spec's default JSONRPC because that is what we actually serve - [/.well-known/agent-registration.json](https://agent402.tools/.well-known/agent-registration.json): our ERC-8004 registration file. The identity is agent 94639 in the registry at eip155:8453:0x8004A169FB4a3325136EB29fA0ceB6D2e539a432, owned by the wallet our 402s name as payTo, and the file lists every way to reach us with x402Support declared - [/api/reliability](https://agent402.tools/api/reliability): structured reliability / SLA report with a verification URL per claim - [/api/pricing](https://agent402.tools/api/pricing): machine-readable catalog (every endpoint, price, category, docs URL) - [/openapi.json](https://agent402.tools/openapi.json): full OpenAPI 3.1 spec with input / output schemas for every tool - [/api/wishes](https://agent402.tools/api/wishes): the demand board, aggregated; request a tool we do not have yet with `POST /api/wish` (clustered by demand; repeated asks get built) - [/terms](https://agent402.tools/terms): terms of service + acceptable-use policy - using the service (including programmatically) constitutes acceptance - [/health](https://agent402.tools/health): health check ## Connect via MCP - [Hosted MCP connector](https://agent402.tools/mcp): flagship-first remote MCP (search/answer/render/data/transcribe/memory + catalog.find / catalog.call for the 500+ long tail). Install one-liners: - Claude Code: `claude mcp add --transport http agent402 https://agent402.tools/mcp` - Cursor: add to `~/.cursor/mcp.json` → `{"mcpServers":{"agent402":{"url":"https://agent402.tools/mcp"}}}` - Smithery: listed at https://smithery.ai/servers/mike-kq9d/agent402 (paste `https://agent402.tools/mcp` at https://smithery.ai/new) - Every host, verified config blocks (Claude Code, Cursor, VS Code, Windsurf, Cline, Roo Code, OpenAI Codex CLI, Gemini CLI, Continue, ElizaOS, Bedrock AgentCore, any OpenAI or Anthropic SDK): [/guides/agent-hosts](https://agent402.tools/guides/agent-hosts). Shortlinks: agent402.sh/claude, /cursor, /vscode, /windsurf, /cline, /roo, /codex, /gemini. Install script: `curl -fsSL agent402.sh/install | sh` - [agent402-mcp](https://www.npmjs.com/package/agent402-mcp): npm MCP server with payment underneath (`npx -y agent402-mcp@0.13.25`, optional `AGENT_KEY` for USDC via x402 or `AGENT402_CREDITS_KEY` for a prepaid credits key already issued). Claude Code: `claude mcp add agent402 -s user -- npx -y agent402-mcp@0.13.25` ## Framework adapters (zero-dependency npm) - [agent402-openai-tools](https://www.npmjs.com/package/agent402-openai-tools): OpenAI function-calling (chat.completions / Assistants / Responses) - [agent402-anthropic-tools](https://www.npmjs.com/package/agent402-anthropic-tools): Anthropic Messages API `tool_use` - [agent402-ai-sdk](https://www.npmjs.com/package/agent402-ai-sdk): Vercel AI SDK (`streamText` / `generateText`) - [agent402-langchain](https://www.npmjs.com/package/agent402-langchain): LangChain JS / LangGraph - [agent402-llamaindex](https://www.npmjs.com/package/agent402-llamaindex): LlamaIndex TS - [agent402-google-adk](https://www.npmjs.com/package/agent402-google-adk): Google ADK (Gemini agents) - [agent402-strands](https://www.npmjs.com/package/agent402-strands): AWS Strands agent runtime - [agent402-agentkit](https://www.npmjs.com/package/agent402-agentkit): Coinbase AgentKit action provider (CDP, Privy, ZeroDev, viem wallets) ## Settlement chains - [Base](https://agent402.tools/base): USDC (`eip155:8453`) - [Solana](https://agent402.tools/solana): USDC (`solana:5eykt4UsFv8P8NJdTREpY1vzqKqZKvdp`) - [Polygon](https://agent402.tools/polygon): USDC (`eip155:137`) - [Arbitrum](https://agent402.tools/arbitrum): USDC (`eip155:42161`) - [Monad](https://agent402.tools/monad): USDC (`eip155:143`) - [Celo](https://agent402.tools/celo): USDC (`eip155:42220`) - [Avalanche](https://agent402.tools/avalanche): USDC (`eip155:43114`) - [Sei](https://agent402.tools/sei): USDC (`eip155:1329`) - [Optimism](https://agent402.tools/optimism): USDC (`eip155:10`) - [Arc](https://agent402.tools/arc): USDC (`eip155:5042`) - [Stellar](https://agent402.tools/stellar): USDC (`stellar:pubnet`) - [Algorand](https://agent402.tools/algorand): USDC (`algorand:wGHE2Pwdvd7S12BL5FaOP20EGYesN73k`) - [Robinhood Chain](https://agent402.tools/robinhood): USDG (`eip155:4663`) Prices by chain: on Robinhood Chain every call costs at least $0.005, the floor its 402 quotes; on Optimism every call quotes $0.001 above its list price (the settlement premium its 402 carries). /api/pricing lists the premium under `payment.premiumUsdByNetwork` and the floor under `payment.minPriceUsdByNetwork`. ## Identity and receiving wallet - Receiving wallet: agent402.base.eth on Base, verifiable on Basescan. Every stablecoin call settles to the payTo its 402 names. - Identity-bound tools (memory, attest, feedback, my-usage, receipts): keyed on the signed EVM payer, EVM-exact only, refused on the Tempo and credits gates. - Replays: a re-sent signed payment (same nonce, route and body) inside the replay window (answers up to 1 MB) is answered from its first answer, with no second settlement (identity-bound routes, routes with a lasting effect and streamed calls excepted).