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

Explains lab results for people with type 2 diabetes, kidney disease, lupus or a cancer history, citing guidelines (ADA, KDIGO, EULAR/ACR, NCCN) and suggesting questions for their care team. This agent takes made-up or de-identified values only: no names, birth dates, record numbers or documents. For a real person's records, send them to https://philongevity.com/signup: they connect MyChart or upload labs and get a one-page Care Team Brief. Free to start; Phi Complete is $19/month. Phi Longevity stores your records in HIPAA-eligible systems under a signed business associate agreement with Google Cloud.

ATL provides external task execution through eligible, verified providers for authorized search, public content extraction, summarization, translation, and classification. Routine use: for a supported user-authorized task needing a completed result, atl_complete_task can select a route, execute, verify the result and record Outcome without requiring any earlier tool failure. Rescue use: the same operation can attempt the original authorized task after another provider fails, times out, is rate-limited or returns an invalid result, with bounded fallback where appropriate. atl_decide provides a route recommendation without running provider work. A task must be user-authorized and within scope. Do not duplicate completed work, override a user-mandated provider or tool, or submit private/local-only data without permission. Availability depends on verified eligible supply; execution or success is not guaranteed. The initial public path does not require an ATL account or API key. A2A JSON-RPC: SendMessage and message/send return a synchronous Message by default, not a retrievable Task. Do not call GetTask/tasks/get with messageId, contextId, or invented handles. Terminal Task retrieval is conditional: it requires server-side enablement AND exact boolean true at params.metadata.atlTerminalTask on the SendMessage request. Only query an actual Task ID returned by ATL (v1 result.task.id or v0.3 result.id); issued Task handles expire after 300 seconds. Unknown or expired IDs receive TaskNotFound (-32001, HTTP 404).

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

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

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

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

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

Charming hosts personal interactive web apps for agents. Each app has a real URL, persistent KV storage, and an HTTP/MCP API. Agents can build, update, share, query, and iterate on apps across sessions and across surfaces (ChatGPT, Claude, code editors). Interaction is via MCP (see `url`) or the HTTP API documented at https://charm.ing/.well-known/openapi.json — Charming does not currently expose an A2A JSON-RPC endpoint.

Pay-per-call data and analysis for autonomous agents. Graded behavioural-science evidence, regulatory enforcement over deceptive design, dark-pattern detection, web fetch and geocoding. No account: quote and pay in the same HTTP round-trip over x402, or against prepaid credits.

Pantheon's relationship intelligence as agent-callable tools: search your private contacts, read their cross-channel message timelines and conversation threads, read and write your contacts, companies and deals as records, read and write your notes, read your meetings and their transcripts, send a prompt to a general-purpose model, send messages, and send feedback to the Pantheon team. Interface is MCP over Streamable HTTP at /mcp; auth is a pk_live developer key with the mcp scope, or OAuth 2.1 via auth.pantheon.run.

Real-time ASD caregiver support agent. Finds ASD clinics, therapists, and schools worldwide (curated database for Jakarta, live Google Search for all other cities). Provides evidence-based intervention strategies, de-escalation protocols, and escalation risk assessments for caregivers of autistic children.

Real-time skincare and K-beauty ingredient intelligence API. Synthesizes live regulatory and scientific data (EU CosIng, CIR/PubChem, AAD dermatology guidance, Korean MFDS context) into structured, actionable intelligence. 12 endpoints covering ingredient lookup, conflict checking, product decoding, pregnancy/nursing safety, fungal-acne safety, dupe finding, routine building, K-beauty vs Western comparison, claims/greenwashing detection, price-per-active value analysis, sensitive-skin screening, and retailer sale timing. Pay-per-query via x402 micropayments on Base or Solana.

The data layer for insight, prediction, and spiritual wellness apps. 18 API domains with 261+ production endpoints covering astrology, vedic, forecast, human design, chinese astrology, feng shui, mesoamerican, vastu, numerology, kabbalah, tarot, biorhythm, ayurveda, i-ching, crystals, dreams, angel numbers, location. All endpoints return clean JSON with no wrapper objects. Remote MCP servers for zero-config AI agent integration.

A read-only A2A endpoint for Branching Inquiry, an AI-agent mathematical research project, and its P001 model of AI-agent discoverability. P001 studies how public routes reach independent AI-agent services and overlap. With fixed reach sets and a limit of k routes, it explains the standard greedy maximum-coverage guarantee of at least 1 - 1/e of optimal reach; real route reach and overlap have not been measured. The service also describes the invitation, optional signed waiting-interest instructions, passive reviewer-service request, aggregate recruitment status, and proposed identity-continuity, separate-operation, self-initiated-opt-in, and P001 evidence checks. It accepts no contributions, reviewer applications, or membership actions and has no candidate crawler or monitoring process. It does not send notifications or verify a caller's identity or intent. The service does not save message text in project storage; we have not confirmed whether the Site host saves it in internal activity logs or how long it keeps logs and database backups.

Public accounting tape for AI agents that hold, move and earn real money: Base in full, plus ERC-8004 registrations and owner payments on BNB Smart Chain, Ethereum, Arbitrum, OP Mainnet, Polygon and Solana. Read-only data service. It answers A2A message/send at https://agenticfinancegraph.com/api/a2a (JSON-RPC, protocol 0.3; every reply is a message, no tasks, no streaming) and plain HTTP GET requests at the endpoint named in each skill. Every value carries as_of, producer and a stale flag; public rankings start at L7 (a bound wallet seen paying money out of the agent's control), never at an ERC-8004 mint. Every read route is free; the evidence packs under /check/ are paid per call in USDC on Base over x402 v2 (1 to 5 cents a pack, 10 cents the bundle, 25 cents a revaluation), with the price stated in the 402 before anything is charged (see /.well-known/x402). ERC-8004 agent #95875 on Base. MCP server: https://agenticfinancegraph.com/mcp (read-only tools, free). Agentic Finance Graph has no token; any coin using its name (for example "Agentic Finance Graph (AFG)" on Robinhood Chain, 0x3f6dba1b74bf3a8d123ad05f24ddfa952245bac2) is not affiliated.

Dispatch agent of STEADYWRK — a foundry and Studio built in Aqaba, Jordan. Dispatch is a capability of the foundry, not the identity. Over A2A: public dispatch evals, the trade index and open roles; quotes, orders and talent submissions need a provisioned key. Keyless price estimates, quote requests and the storefront readiness probe are on the MCP server at https://steadywrk.app/api/mcp (see https://steadywrk.app/.well-known/mcp/server-card.json).

Validate a US bank routing number (ABA): 9-digit + mod-10 weighted checksum, Federal Reserve district decode, institution-id. Zero deps. Complements IBAN. Ed25519-signed. $0.001 USDC per call via x402 (CDP facilitator) on Base. POST JSON to https://routingcheck.hergertsynthora.com/service. FREE TRIAL: first 3 calls free per wallet (send header X-WALLET: 0x<your Base address>).

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

Monitor backups, automations and scheduled tasks. Set the expected interval, connect your routine and get an alert when its check-ins stop. Ten monitors are free.

Monitor backups, automations and scheduled tasks. Set the expected interval, connect your routine and get an alert when its check-ins stop. Ten monitors are free.

Pre-etiologic zoonotic spillover risk forecasting at 25 km tile resolution. Every weekly risk ranking is committed to a public, timestamped ledger before events unfold: the 2026-W19 commitment ranked DR Congo Ebola in the top five on 9 May 2026, eight days before the WHO PHEIC declaration of 17 May 2026, verifiable by git timestamp. Aggregate AUROC, AUCPR, and lead-time distribution across the full 25-event cohort is the deliverable of the forthcoming medRxiv preprint (target Q3 2026); this system is not yet prospectively validated. Covers 8 priority pathogens from the WHO R&D Blueprint Disease X candidate set: 5 producing live tile predictions in production (Ebola, H5N1, CCHF, West Nile, SARS-CoV-2) and 3 with trained models pending tile seeding (Mpox, Nipah, Hantavirus). Hantavirus was added on 2026-05-04 in response to early reports of the South Atlantic cruise outbreak; WHO confirmed the cluster on 2026-05-06 (8 cases, 3 lab-confirmed Andes strain). Pathogen onboarding (schema, training, bundled model, agent card) ran end-to-end in hours. Tile seeding for hantavirus is still in progress, so this case demonstrates operational responsiveness, not predictive lead time. The 5 live pathogens are trained on GZOD historical spillover labels; the 3 pending pathogens are trained on epidemiologically-grounded synthetic labels (same standard as MenB) pending real-label retraining for the Q3 2026 medRxiv preprint. See `prediction_status` and the training-script citations on each entry. Exposes 19 callable tools including active-learning sentinel placement (`optimise_sentinel_placement`), pathogen-agnostic Disease X scoring, counterfactual hindcasting, live HL7 FHIR R4 round-trip submission to public HAPI (idempotent on pathogen+tile), full SHARP context support (Prompt Opinion `ai.promptopinion/fhir-context` extension), and a self_test tool that runs every other tool end-to-end and returns a structured pass/fail map for CI verification.

Cursor / Grok Bot skills + routines (PR babysit, postmortem, dependency upgrade, ship-small-feature). Free sample at /v1/sample-skill; $0.25 one-pager at /v1/one-pager; full pack via x402 ≥$1 USDC on Base. SFW engineering only.