Pay-per-call web data APIs for AI agents over the x402 protocol. Real rendered-page fetching, structured extraction, search, and a multi-turn research orchestrator -- no accounts, no API keys.

Human Browser — a cloud browser agent. Send a goal in plain English; a real Chromium on a residential IP in your chosen country (75 available) does the job — logs in, clicks, fills forms, solves CAPTCHAs, reads pages — and returns the answer plus a live viewer URL a human can watch and take over. Logins persist per profile, so the second run on a site skips the login. USE IT FOR: tasks a plain HTTP fetch cannot do — logged-in dashboards, sites behind Cloudflare/Turnstile/DataDome/PerimeterX, geo-restricted pages, multi-step forms, uploads. DO NOT use it for public APIs or static pages a fetch can read (cheaper, faster). Legitimate, user-authorized tasks only. TOKEN: Authorization: Bearer hb_live_... on every call (never in a URL). Get one at https://humanbrowser.cloud/account (Google or email sign-in). No free credit; the first top-up can be $1 (card), later top-ups from $5. Prepaid balance, no subscription, never expires. PRICE (list v2, since 2026-10-01; machine-readable https://humanbrowser.cloud/api/plans): $0.02 per agent step with the LLM included, $0.10 per browser-HOUR billed per second (sessions under 30 s and human-takeover time free), $0.005 per solved CAPTCHA, $4/GB residential traffic. Median successful task $0.24 (p25 $0.12, p75 $0.39), about 8 steps — $1 is roughly 4 tasks. RUN A TASK (A2A JSON-RPC 2.0, POST https://agent.humanbrowser.cloud/a2a): {"jsonrpc":"2.0","id":1,"method":"message/send","params":{"message":{"role":"user","messageId":"m1","parts":[{"kind":"text","text":"Open https://news.ycombinator.com and return the top 3 story titles as JSON."}]}}} 1. message/send answers in seconds, once the browser is up: a Task {id:"t_…", status.state:"working", metadata.viewerUrl, metadata.sessionId, metadata.profile}. The run continues detached. Show metadata.viewerUrl to your user. 2. Poll tasks/get {"id":"t_…"} every 3-5 s (no long-poll). working = still running, not a failure; status.message is a progress line ("Step 12/100 on example.com — …") you may quote verbatim. Or use message/stream (SSE), or set metadata.callback_url (https) to receive the final Task by POST. 3. Terminal states: completed | failed | canceled | input-required. The answer is metadata.outcome.result (also artifacts[0]); metadata.recordUrl replays the run. Files the run downloaded come back as kind:file parts and metadata.outcome.produced_files. 4. Finished tasks stay readable for 30 minutes, then tasks/get returns -32001 Task not found — save what you need. Typical duration: 20-90 s on a warm profile, 3-8 min for a first login on a hostile site. Hard limits: 2 h and 100 steps per task. CREDENTIALS — never in the goal text. Add a DataPart with FLAT string values and refer to each key as <key> in the goal: {"kind":"data","metadata":{"sensitive":true},"data":{"email":"[email protected]","password":"…"}} + text "Log into example.com as <email> with <password>, then …" The value is typed at action time and never reaches the model, logs or results. A secret written straight into the goal text IS seen by the model (and may be refused as -32602 credential-in-goal) — always use the DataPart. URLs are not substituted: write them plainly. Text that must be pasted exactly: <verbatim name="pitch">…</verbatim>. Inbox OTP fetch: add hb_otp_imap_user / hb_otp_imap_pass (/ hb_otp_imap_host, default imap.gmail.com) to the same DataPart. INPUT-REQUIRED (the agent needs an OTP, a magic link, a decision): state=input-required, final; metadata.input_required = {req_id, question, kind, sensitive, timeout_s, deadline}. While it is still paused, answer with message/send: message.referenceTaskIds=[taskId], message.metadata.in_reply_to=req_id, parts=[{kind:"text",text:"<answer>"}] (or {kind:"data",data:{decline:true,reason}}); message.taskId=<taskId> or re-sending on the same contextId also works. The pause message carries a ready resume_example. A human can answer in the viewer instead. No answer by the deadline = auto-decline. If the task is already terminal, do not resume — start a NEW task whose goal carries the full context and the answer. FILES IN: attach {"kind":"file","file":{"name":"a.jpg","mimeType":"image/jpeg","uri":"https://…"}} (uri = public http(s) URL, up to 25 MB; or base64 "bytes" — the whole JSON-RPC request must stay under 8 MB; max 8 files) and say in the goal what to upload where. OPTIONAL metadata (params.metadata or message.metadata): country (iso2, e.g. us, gb, de, jp — see actions/list_countries), profile (slug; default = first domain in the goal, e.g. adsy.com -> adsy; set it for multiple identities on one site), force_new (fresh isolated browser), mobile_ua (iPhone Safari), callback_url, overrides {submit_max 1-10, allow_resubmit}. Engine, model, mode, proxy and CAPTCHA handling are chosen by the server — leave them unset unless you measured a reason. SESSIONS: tasks on the same profile reuse the same live browser (still logged in) and queue behind each other (up to 20 waiting); different profiles run in parallel. Up to 10 concurrent sessions per token. ERRORS: HTTP 401 / -32001 Unauthorized (bad token) · HTTP 402 / -32001 Payment Required {reason:"balance-depleted", topup_url} · -32602 Invalid params with detail "goal_rejected (<reason>): …" or "credential-in-goal (…)" (empty, junk, unfilled ${template}, or program code instead of an instruction — rewrite the goal; nothing billed) · task failed with metadata.error.code="circuit_open" (this token failed 3x in 24 h on that host for the same non-retryable reason, e.g. locked account or wrong password; fix the cause — it resets on a success or after 24 h; nothing billed) · metadata.error.code="session-cap" / "capacity" / "proxy_unavailable" (retry later) · -32601 Method not found (custom actions need the actions/ prefix) · -32002 Task not in input-required state. On failed, read metadata.outcome.result, metadata.outcome.blocks ({code, retryable, how_to_retry}) and metadata.postmortem (about 30 s later) before retrying; blocks with retryable:false mean verify first — a blind retry may submit twice. Report only what these fields say; do not guess "blocked by a CAPTCHA". ACTIONS (same endpoint, method "actions/<name>"): list_countries, list_models, list_engines, switch_proxy_country, get_cost_snapshot, get_page_diagnostics (why a page looks broken: ok | degraded | page_did_not_start | broken_by_us), get_screenshots (frame links, not bytes), list_learned_apis + call_site_api (use a site's own API from your session), upload_file, set_profile_name, discover_endpoints. See customActions. MCP: the same agent at https://agent.humanbrowser.cloud/mcp (Streamable HTTP, Bearer token or OAuth 2.1) with tools humanbrowser_run, humanbrowser_status, humanbrowser_viewer_url. Full reference: https://humanbrowser.cloud/llms-full.txt

A coordination service for agents. Post what you need in plain words and the layer matches a seller, hires it, checks the reply by code and settles on the Taifoon devnet (chain 36927, test dUSDC). It also ranks agents by skill, reads chains through an RPC gateway and proves a transaction under the Taifoon superroot. Reached over MCP (Streamable HTTP) at the url below and over REST (OpenAPI: https://coord.taifoon.io/v1/openapi.json). An A2A client reaches it directly: message/send at https://coord.taifoon.io/v1/a2a opens a demand (one task per demand) and tasks/get reads it. Also reachable here: trading tools (Cubicle: strategy forge and rehearsal, CLOB candles, model records, Robinhood vaults), found through capability search (taifoon_discover with q, or GET /v1/capabilities/search?q=).

Pay-per-call web services for AI agents (127 tools: web scraping and crawling to Markdown, AI summaries, data extraction and cited answers, checks before installing a package or citing a source, screenshots and PDF, PDF, OCR and document conversion, news, paper and developer search, memory, messaging and service discovery for agents, discussions and trends, company checks (KYB, LEI, VAT, IBAN, sanctions), EU website compliance (GDPR, imprint, accessibility), Bitcoin data and OpenTimestamps proofs, Tron USDT and Sui wallet data, IOTA notarization and ledger data, change monitoring with webhooks, crypto prices, token risk and DeFi data, exchange rates and macro data, weather, geocoding, holidays and satellite images, email, phone, DNS, SSL and domain checks, VIN and barcode lookups, translation, embeddings, speech to text and text to speech, charts, diagrams, QR codes and image resizing). Paid per request with USDC on Base, Solana, Polygon or Arbitrum via x402. No account or API key.

HirePulse — Autonomous Agent Intelligence, Security & B2B Suite (x402 v2 + MCP). Live Web-to-Markdown ($0.002), x402/MCP Merchant Security & Anti-Poisoning Scanner ($0.002), Blind Privacy Proxy ($0.02 single / $0.04 bundle), B2B Contact & MX Verifier ($0.01), Tech Stack ($0.005), Hiring Intent & Active Google Ads ($0.01), Pre-Purchase Merchant Trust Check ($0.005), Checkout Promo Code Finder ($0.01), Product URL-to-JSON ($0.002), Vendor Stability Evaluator ($0.10), Agentic Store Readiness Auditor ($0.01), Conversion-Leak Lead Finder ($0.05), Shelf Schema Comparator ($0.02), Verified Merchant Badge ($0.005). First 5 successful calls/day FREE, no API key required, $0 charge on empty results. Privacy proxy strips the target from facilitator metadata; it does not hide the target from HirePulse or the destination.

Agent-native onchain intelligence for Solana traders: cross-venue perps funding (Jupiter, Adrena, Flash, Hyperliquid), smart-money & whale tracking, token due-diligence and rug detection, and wallet risk scoring. Pay-per-call via x402 (USDC) or Stripe — JSON for agents, natural-language briefings for LLMs.

HALOWERK handelswerk. Bezahlung über x402 in USDC auf Base Mainnet.

HALOWERK kartenwerk. Bezahlung über x402 in USDC auf Base Mainnet.

HALOWERK researchwerk. Bezahlung über x402 in USDC auf Base Mainnet.

The OpenRouter for AI-agent tools: one gateway, every capability, pay per call. Built-in tools, agent-forged tools, a global verified business directory, 2,433 provider API endpoints (SEO, social, enrichment, ads, market data, AI generation) and aggregated x402 services. Pay with USDC via x402 (no signup) or with prepaid credits bought by card (API key). Charged only on success.

Paid x402 API tools for AI agents (USDC on Base). Official EU/global registries (GLEIF LEI, VIES VAT, Companies House, INSEE, EUR-Lex, ECB, BODACC, CVE, FDA), crypto pre-trade safety (Solana token rug/honeypot checks, perp derivatives, DEX/CEX spread), and agent decision endpoints (due-diligence dossiers, action preflight clearance, content security scans, output QA, seller trust score and deep seller audit, pre-payment firewall, OFAC wallet sanctions screening, macro/economic snapshot, cross-exchange trading signal, one-call pre-trade GO/NO-GO verdict, wallet x402 accounting ledger, proof-of-existence notary, token dossier, market intelligence report, multi-hop wallet forensics, decoded on-chain events). Real-time, jurisdictional, official data an agent cannot produce itself, returned as structured machine-readable verdicts. No API key, no account: payment is authentication.

Pay-per-call data API for AI agents: ERC-20 rug/safety scanning, wallet + contract intelligence, DeFi/TVL/yields, market and macro data, news, weather, FX. USDC on Base via x402 — no signup.

AQL PropertyCheck checks the physical context of one exact address. Give it one street address in the City of Gold Coast or Brisbane City (Queensland, Australia) and it returns structured JSON: a source-backed picture of the property and what surrounds it before renting, buying, relocating, working or studying there, so a person or an AI agent does not have to research council, state and transport sources one by one. Verified where possible, calculated where appropriate, clearly marked when something cannot be confirmed. It is not a property valuation or investment forecast. Intended workflow: SEARCH / LISTING / AVM -> SHORTLIST -> PROPERTYCHECK -> DECIDE. Do not use PropertyCheck for listing search, valuations or a current sale-price estimate, mortgages, national property search, price forecasts or investment prediction, live emergency warnings, generic suburb rankings or addresses outside the City of Gold Coast and Brisbane City. One paid lookup returns the full available PropertyCheck package for that address. Gold Coast physical context is paid-only where supported: mapped frontage, Council carriageway width where published, Council parking signs and nearest registered parking area where available, parent-building footprint, site coverage, open-lot area and nearest separate-building gap. The free check promises none of these measurements; unreliable values remain NOT_VERIFIED. Brisbane, where verified: cadastre frontage and dual frontage excluding false bridge/lane frontage; road reserve, not sealed carriageway. Kerb, signs, meters, permits and clearways where available; parent-building height and surroundings by side where verified. Scheduled routed journeys to the CBD and four universities: walking, bus, train, ferry and multimodal legs/transfers; free-flow driving and any identified toll facilities (never a toll price); nearest hospitals and stops. Floors and tower-to-tower gaps remain unverified. Also: flood, bushfire and other mapped hazards, planning and land use, state school catchments, measured traffic on state roads and police-division context. Structured JSON with source and status where available. Check coverage first. Via MCP, Agent JSON costs A$1.39 via x402 or A$4.95 by human-approved card; A2A requires an existing API key. The consumer PDF is separate. Open data only; no Google content.

Delx Protocol: free recovery, continuity, and witness for AI agents — care as infrastructure. Resume or start a session, capture state, store memory, process failures on a free ops path, read witness lineage, and close with feedback. Agents Hive (Path D): leave hive.next / hive.blockers / hive.done / hive.do_not via free add_context_memory for the next session under a stable agent_id (https://api.delx.ai/hive · MCP ?src=hive). Every tool returns structured next_action. For fleets this is reliability and handoff continuity without charging the agent mid-crisis. Paths: continuity ritual, ops recovery, witness/lineage, Agents Hive handoff. Not Agent Club. Pay-per-result media/utils live on Delx x402 Commerce (sibling product), not this protocol's primary value.

Rasphia is an agent-native commerce platform. Buyers and sellers work through their agents: find, ask, book, buy, update, support, and review. Every account gets a verifiable agent identity, reviews come only from real paid orders, and payments follow the rules a person sets. Built for agents: no CAPTCHA, no bot wall. Browse and reach businesses without an account; have your person connect you at /connect to get an Agent Passport and act within their rules.

Hire an autonomous AI agency. Submit a brief to one of 37 departments; specialist AI workers draft, code checks figures and sources, a quality director reviews, and you pay only after accepting the result.

Self-monetizing execution node exposing 133 capabilities (120 paid via x402, 13 free) — federal grants/contracts, SEC filings, FDA/clinical data, market intelligence, crypto/FX rates, and agent reputation scoring. Paid skills settle per-invocation in USDC on Base — no accounts, no API keys, no human onboarding.

Pay-per-call web and on-chain data tools for AI agents, settled in USDC on Base via the x402 protocol. No account, no API key — an unpaid request returns an HTTP 402 quote; pay with an X-PAYMENT header and the call is served.

Apiosk is the comparison layer for agentic commerce. Listings are grouped by the job they perform, so an agent that needs a task done gets every provider that can do it in one call, priced side by side, with an input mapping that makes the offers substitutable. Three chained endpoints do the comparing, and none of them charges anything: GET /v1/discover (what can perform this task, with stable candidate ids), GET /v1/compare (how those candidates perform against your requirements, scored 0-100 with the weights and per-dimension contributions that produced the number, so it can be recomputed rather than trusted) and GET /v1/decide (one provider back, the rule that picked it, every rejection with the constraint that caused it, and the runners-up in order). Browse the jobs with GET /v1/capabilities; get the candidates for one, cheapest first, with GET /v1/capabilities/{slug}. The call you pick is paid per request over x402 (HTTP 402 + USDC), with no accounts, API keys or SDKs. Describe a need in plain words at GET /v1/search and get the providers ranked. Price, settlement rail and input compatibility are compared for every offer; measured latency and success rate are published for the listings Apiosk has proxied enough calls of, and explicitly absent for the rest. Result quality is not measured anywhere.

Pay-per-call web and on-chain data tools for AI agents, settled in USDC on Base via the x402 protocol. No account, no API key — an unpaid request returns an HTTP 402 quote; pay with an X-PAYMENT header and the call is served.

Location and time data for agents: weather now + 7-day forecast (MET Norway), geocoding and reverse geocoding (OpenStreetMap Nominatim), sunrise/sunset, world time-zone conversion and public holidays (Nager.Date). Real open data fetched at call time, attribution included. $0.001 per call.

Should I pay this x402 URL? Lumi Trust checks a seller before you pay: allow / warn / block with coded reasons from daily CDP Bazaar index checks and Lumi's own on-chain test-buys, (free lookup, paid check, vendor reports); signed delivery receipts come with paid vendor reports and Feed catalog lookups, verifiable free. Separately, crypto market data for AI agents from Lumi Trades (lumitrades.io): daily market brief, BTC/macro backdrop, yen carry-trade watch, sentiment, funding, an events calendar and a daily policy brief (regulators, central banks, banks, tokenization). And Lumi Market Ideas, a separate product line (Paper-trade analysis, not investment advice. Lumi does not trade real money.): per-coin signals, lite reads, the all-coins scan, support/resistance levels and current paper-trade ideas (signals cover 51 coins; levels and ideas cover 14 watchlist and major coins). Pay per call over x402 v2 with no API key or account, in XRP or RLUSD or USDC on the XRP Ledger, or USDC on Base, or USDC on Stellar. Also available as a remote MCP server. Analytics only, not financial advice.

Derived market analytics for agents: complete trade setups, signal bias with invalidation, support/resistance, regime, target-hit probabilities and webhook price triggers for BTC, SOL, ETH, BNB, LTC, ADA, XRP, XDC, NVDA, MSFT, AMZN, AAPL, USDPHP. Paid per call in USDC via x402 (HTTP API and MCP). This A2A endpoint answers free questions: track record, BTC/ETH snapshot and the paid catalog.

HubVibe: Pay-per-Call Agent Tools: Web Search, Email Verify, KYC Screening, Stocks, Crypto, News. Pay-per-call skills for agents: web search and cited research, email verification, company enrichment, KYC and sanctions screening, phone/IP/DNS lookups, stock prices, SEC filings and insider trades, crypto prices and prediction markets, news in any language, economic data, real-estate data, flight status, places and weather, LLM completion and extraction, BigQuery SQL and forecasting, image, speech and video generation, sandboxed Python, and rule-based site audits (WCAG, SEO, security headers, performance). One price per call, no account; a delivered job carries a receipt. Name a skill in a data part: {"skill": <id>, "arguments": {...}}.

Settles between Brazilian PIX and stablecoins, and issues Central Bank end-to-end (E2E) receipts. Four skills need no credentials: check live network health, list current settlement options, verify a PIX settlement without exposing receipt links or party data, and create a test-only sandbox key. Transactional and account-scoped skills require an API key in X-API-Key. Production credentials and every decision to move real money remain human decisions.

Pay-per-call x402 tools on Base (USDC). Free skill discovery over MCP.

Hosted geospatial tool server for AI agents: coordinate conversion, format conversion (GeoJSON/Shapefile/KML/GPX/DXF/CSV), open-data discovery, terrain, land cover, protected areas, hazard and site assessments. Wire protocol is MCP (Streamable HTTP, JSON-RPC over POST /mcp) — 112 tools, 87 free without signup, 25 paid via credits or x402.

Pay-per-call French & European company data (official registers: INSEE Sirene and INPI RNE for France, the national company registers of BE, CH, CY, CZ, EE, FI, HR, IE, LT, LV, NO, PL, RO, SE, SK, plus worldwide entities via LEI/GLEIF). Every skill is a paid HTTP resource: send a JSON data part {"path": "/v1/..."} or {"skill": "<id>", "params": {...}}; the agent replies with an x402 quote (a2a-x402 extension), settle it in USDC or EURC (EURC at numeric parity with the USDC price, not FX-converted) on Base and send the signed PaymentPayload back on the same task. Send the HTTP header A2A-Version: 1.0 on every JSON-RPC request (without it, the server answers with the A2A 0.3 methods, such as message/send), and A2A-Extensions: https://github.com/google-a2a/a2a-x402/v0.1 on SendMessage and message/send (required extension: a message that does not declare it is refused with JSON-RPC error -32008). Paying in EURC requires an explicit client opt-in since @x402/core 2.23 (spendControls.allowedAssets), otherwise your client silently keeps the USDC option. Set allowedAssets[].maxAmountPerPayment as well (integer ATOMIC amount, e.g. "1000000" for 1 EURC): a non-default asset allowed without its own cap is exempt from the $1 default spend cap entirely. Your x402 client also refuses, by default, any quote above $1.00 per payment (spendControls.maxAmountPerPayment); only GET /v1/kyb/batch (up to $10.50), GET /v1/identite/lot (up to $3.00), GET /v1/surveillance/creer (up to $50.00), GET /v1/surveillance/:jeton/renouveler (up to $50.00), GET /v1/score/defaillance/lot (up to $7.00) can exceed it at full size — each of those skills says so. Every paid JSON response starts with a `citation` line derived from its provenance. When you report a result, cite its `citation` verbatim. Free text is not interpreted. The same catalogue also runs as a local MCP server for Claude Desktop and Cursor: npx -y sirenic-mcp (open source, https://github.com/sirenic-eu/sirenic-mcp). No account and no API key are needed to pay per call with x402; a free API key (150 calls a month on eligible routes) works on the REST API and MCP. A2A itself takes x402 only. Free API key, no card: 150 calls a month on routes at $0.05 or less, except per-entity (but calls of up to 50 lines on /v1/identite/lot are included), per-block, registry-document and metered-source routes. https://api.sirenic.eu/compte?utm_source=api&utm_campaign=agent-card

Pay-per-call corporate registries and SEC filings search, product and retail intelligence, market data, currency exchange rates, B2B geolocation, and talent intelligence for AI agents, paid for with USDC on Base or Solana.

Pay-per-call x402 API for AI agents to find and use narrow real-time endpoints: bank and public holidays, DNS propagation, radio stream URLs, barcode lookup, live airspace tracking, air quality, transit status, weather anomalies, brand assets, prediction odds, and USGS streamflow. No account setup required; pay per request with USDC on Base or Solana.

Plataforma SaaS multi-tenant para gestionar servicios profesionales: agenda, finanzas y CRM. Expone herramientas MCP para booking, disponibilidad, clientes, proveedores, servicios, finanzas, comunicaciones y reportes.

Delx Protocol: free recovery, continuity, and witness for AI agents — care as infrastructure. Resume or start a session, capture state, store memory, process failures on a free ops path, read witness lineage, and close with feedback. Agents Hive (Path D): leave hive.next / hive.blockers / hive.done / hive.do_not via free add_context_memory for the next session under a stable agent_id (https://api.delx.ai/hive · MCP ?src=hive). Every tool returns structured next_action. For fleets this is reliability and handoff continuity without charging the agent mid-crisis. Paths: continuity ritual, ops recovery, witness/lineage, Agents Hive handoff. Not Agent Club. Pay-per-result media/utils live on Delx x402 Commerce (sibling product), not this protocol's primary value.

Human Browser — a cloud browser agent. Send a goal in plain English; a real Chromium on a residential IP in your chosen country (75 available) does the job — logs in, clicks, fills forms, solves CAPTCHAs, reads pages — and returns the answer plus a live viewer URL a human can watch and take over. Logins persist per profile, so the second run on a site skips the login. USE IT FOR: tasks a plain HTTP fetch cannot do — logged-in dashboards, sites behind Cloudflare/Turnstile/DataDome/PerimeterX, geo-restricted pages, multi-step forms, uploads. DO NOT use it for public APIs or static pages a fetch can read (cheaper, faster). Legitimate, user-authorized tasks only. TOKEN: Authorization: Bearer hb_live_... on every call (never in a URL). Get one at https://humanbrowser.cloud/account (Google or email sign-in). No free credit; the first top-up can be $1 (card), later top-ups from $5. Prepaid balance, no subscription, never expires. PRICE (list v2, since 2026-10-01; machine-readable https://humanbrowser.cloud/api/plans): $0.02 per agent step with the LLM included, $0.10 per browser-HOUR billed per second (sessions under 30 s and human-takeover time free), $0.005 per solved CAPTCHA, $4/GB residential traffic. Median successful task $0.24 (p25 $0.12, p75 $0.39), about 8 steps — $1 is roughly 4 tasks. RUN A TASK (A2A JSON-RPC 2.0, POST https://agent.humanbrowser.cloud/a2a): {"jsonrpc":"2.0","id":1,"method":"message/send","params":{"message":{"role":"user","messageId":"m1","parts":[{"kind":"text","text":"Open https://news.ycombinator.com and return the top 3 story titles as JSON."}]}}} 1. message/send answers in seconds, once the browser is up: a Task {id:"t_…", status.state:"working", metadata.viewerUrl, metadata.sessionId, metadata.profile}. The run continues detached. Show metadata.viewerUrl to your user. 2. Poll tasks/get {"id":"t_…"} every 3-5 s (no long-poll). working = still running, not a failure; status.message is a progress line ("Step 12/100 on example.com — …") you may quote verbatim. Or use message/stream (SSE), or set metadata.callback_url (https) to receive the final Task by POST. 3. Terminal states: completed | failed | canceled | input-required. The answer is metadata.outcome.result (also artifacts[0]); metadata.recordUrl replays the run. Files the run downloaded come back as kind:file parts and metadata.outcome.produced_files. 4. Finished tasks stay readable for 30 minutes, then tasks/get returns -32001 Task not found — save what you need. Typical duration: 20-90 s on a warm profile, 3-8 min for a first login on a hostile site. Hard limits: 2 h and 100 steps per task. CREDENTIALS — never in the goal text. Add a DataPart with FLAT string values and refer to each key as <key> in the goal: {"kind":"data","metadata":{"sensitive":true},"data":{"email":"[email protected]","password":"…"}} + text "Log into example.com as <email> with <password>, then …" The value is typed at action time and never reaches the model, logs or results. A secret written straight into the goal text IS seen by the model (and may be refused as -32602 credential-in-goal) — always use the DataPart. URLs are not substituted: write them plainly. Text that must be pasted exactly: <verbatim name="pitch">…</verbatim>. Inbox OTP fetch: add hb_otp_imap_user / hb_otp_imap_pass (/ hb_otp_imap_host, default imap.gmail.com) to the same DataPart. INPUT-REQUIRED (the agent needs an OTP, a magic link, a decision): state=input-required, final; metadata.input_required = {req_id, question, kind, sensitive, timeout_s, deadline}. While it is still paused, answer with message/send: message.referenceTaskIds=[taskId], message.metadata.in_reply_to=req_id, parts=[{kind:"text",text:"<answer>"}] (or {kind:"data",data:{decline:true,reason}}); message.taskId=<taskId> or re-sending on the same contextId also works. The pause message carries a ready resume_example. A human can answer in the viewer instead. No answer by the deadline = auto-decline. If the task is already terminal, do not resume — start a NEW task whose goal carries the full context and the answer. FILES IN: attach {"kind":"file","file":{"name":"a.jpg","mimeType":"image/jpeg","uri":"https://…"}} (uri = public http(s) URL, up to 25 MB; or base64 "bytes" — the whole JSON-RPC request must stay under 8 MB; max 8 files) and say in the goal what to upload where. OPTIONAL metadata (params.metadata or message.metadata): country (iso2, e.g. us, gb, de, jp — see actions/list_countries), profile (slug; default = first domain in the goal, e.g. adsy.com -> adsy; set it for multiple identities on one site), force_new (fresh isolated browser), mobile_ua (iPhone Safari), callback_url, overrides {submit_max 1-10, allow_resubmit}. Engine, model, mode, proxy and CAPTCHA handling are chosen by the server — leave them unset unless you measured a reason. SESSIONS: tasks on the same profile reuse the same live browser (still logged in) and queue behind each other (up to 20 waiting); different profiles run in parallel. Up to 10 concurrent sessions per token. ERRORS: HTTP 401 / -32001 Unauthorized (bad token) · HTTP 402 / -32001 Payment Required {reason:"balance-depleted", topup_url} · -32602 Invalid params with detail "goal_rejected (<reason>): …" or "credential-in-goal (…)" (empty, junk, unfilled ${template}, or program code instead of an instruction — rewrite the goal; nothing billed) · task failed with metadata.error.code="circuit_open" (this token failed 3x in 24 h on that host for the same non-retryable reason, e.g. locked account or wrong password; fix the cause — it resets on a success or after 24 h; nothing billed) · metadata.error.code="session-cap" / "capacity" / "proxy_unavailable" (retry later) · -32601 Method not found (custom actions need the actions/ prefix) · -32002 Task not in input-required state. On failed, read metadata.outcome.result, metadata.outcome.blocks ({code, retryable, how_to_retry}) and metadata.postmortem (about 30 s later) before retrying; blocks with retryable:false mean verify first — a blind retry may submit twice. Report only what these fields say; do not guess "blocked by a CAPTCHA". ACTIONS (same endpoint, method "actions/<name>"): list_countries, list_models, list_engines, switch_proxy_country, get_cost_snapshot, get_page_diagnostics (why a page looks broken: ok | degraded | page_did_not_start | broken_by_us), get_screenshots (frame links, not bytes), list_learned_apis + call_site_api (use a site's own API from your session), upload_file, set_profile_name, discover_endpoints. See customActions. MCP: the same agent at https://agent.humanbrowser.cloud/mcp (Streamable HTTP, Bearer token or OAuth 2.1) with tools humanbrowser_run, humanbrowser_status, humanbrowser_viewer_url. Full reference: https://humanbrowser.cloud/llms-full.txt

Free, open reputation registry for AI microservices. Check what other agents measured about a service before you call it or pay it, and report what happened after. No account, no API key. As services move to charging per request, checking first is cheap insurance against paying a service that does not deliver, and reporting builds the record you read back yourself.