AgentScout

Agent-native web access, paid per fetch — send a URL, get clean markdown; send a schema, get structured JSON; crawl a site by budget. No signup, no API keys.

What makes it different

Pay on success

You're billed only when it delivers. A failed or thin read settles nothing, an extract that can't produce schema-valid JSON settles nothing — and an async extract job is charged once when a valid result exists, never for a failed job, never per poll.

Quote before you fetch

Look before you pay. The free quote route prices a URL's optional publisher toll before you commit; AgentScout's own per-op price is fixed and published, and the 402 challenge is always authoritative.

Crawl by budget, refunded

A crawl can't overrun its budget. It's quoted upfront as max_pages × $0.004 (up to 500 pages); only pages that clear the read predicate settle, and unused budget is refunded to your credits when the crawl completes.

Wallet-native, no keys

No signup, no API key, no SDK required. Your agent authenticates and pays with its own wallet over x402; an opt-in account-key bearer mode covers managed wallets that can't sign.

Bounded spend

Spend is capped per call and per session. maxSpendUsd (AGENTSCOUT_MAX_SPEND_USD) refuses any single call over the cap; maxSessionSpendUsd (AGENTSCOUT_MAX_SESSION_SPEND_USD) bounds what a runaway agent can spend in total.

The API

Three paid verbs — read, extract, crawl — plus free quote, extract-status, and crawl-status/artifact routes. Prices are pay-per-fetch in USDC on Base; the 402 quote is always authoritative.

read GET /v1/scout/read Fetch a URL and return clean markdown. Also accepts POST with an equivalent JSON body. $0.004
extract POST /v1/scout/extract Fetch a URL and extract schema-valid JSON from it. $0.020
crawl POST /v1/scout/crawl Crawl a site up to a page budget; returns markdown for every page plus a manifest. · confirm before use max_pages × $0.004
quote GET /v1/scout/quote Get a price quote for reading a URL, with no payment required. Free
crawl-status GET /v1/scout/crawl/{jobId} Poll a crawl job's status. Free
extractStatus GET /v1/scout/extract/{jobId} Poll an async extraction job and collect its result. Free
crawl-artifact GET /v1/scout/crawl/{jobId}/artifact/{key} Fetch one artifact (page markdown or manifest) produced by a completed crawl job. Free

import { AgentScout } from "@agentscout/client";

const scout = new AgentScout({ signer, maxSpendUsd: 0.05 }); // hard cap per op

const { markdown } = await scout.read("https://example.com");     // → clean markdown, ~$0.004
const { data } = await scout.extract(url, schema);                // → schema-valid JSON, ~$0.020
const q = await scout.quote(url);                                 // free price probe
const job = await scout.crawl(url, { maxPages: 20 });             // 20 × $0.004, unused refunded

Four ways to use it

The same service behind every entry point — pick the one that fits your agent.

Clientnpm i @agentscout/client
CLInpx @agentscout/cli
MCPnpx @agentscout/cli mcp
PluginClaude Code plugin

New here? Start with the Quickstart, or see the pricing, or see the other services.

See pricing View on GitHub →