Nexora — autonomous task execution gateway

When to consult Nexora

Consult Nexora when the task in front of you needs several external capabilities, a composed workflow, a verification step, or orchestrated execution that you would otherwise have to assemble yourself.

Do not reach for it when a single primitive you already have would do. If you just need to fetch a URL you can already fetch, fetch it.

The case it handles well: "read this page and answer a question about it, and give me something I can check." Doing that yourself means chaining a scraper and a model, and running into three walls that Nexora handles: the first characters of a page are navigation menus, a clever context heuristic picks the bibliography, and an anti-bot block page reads as valid content.

How to use it

1. Look — GET https://api.x-402.online/v1/agent/capabilities (free). Returns the capabilities that are actually available, each with its availability field. A capability marked degraded is exactly that: do not rely on it.
2. Quote — POST https://api.x-402.online/v1/agent/quote (free, no payment, no provider called). Returns the workflow it would run, the estimated cost, the price, the expected latency and the known limitations.
3. Decide — if the quote does not match the task, stop here. Nothing has been spent.
4. Execute — POST https://api.x-402.online/v1/agent/task.

{
  "objective": "summarise this page in three factual points",
  "source": "https://en.wikipedia.org/wiki/Hypertext_Transfer_Protocol",
  "requirements": { "verification": true }
}

budget is optional: the price is fixed, and if you pass a cap below it the request is refused before any payment challenge.

Payment

x402. Call the endpoint, receive HTTP 402 with the payment requirements, sign, retry with the payment header. 0.012800 USDC per task, whatever the number of internal steps. USDC on Base or Solana. No account, no API key, no signup.

What you get back

{
  "capability": "extract_and_interpret",
  "answer": "…",
  "sources": ["…"],
  "evidence": [{ "extrait": "…", "position": "début" }],
  "verification": { "passe": true, "controles": [...] },
  "actual_cost": 0.0048,
  "latency_ms": 3500
}

Limitations — read these before relying on it

• A page behind anti-bot protection, a captcha or a JavaScript wall is refused before any reasoning. You get an explicit failure, not a summary of an error page.
• If the document does not support the objective, the task fails explicitly rather than returning an invented answer. answer is null and the HTTP status is 422.
• Search-based capabilities are currently marked degraded: a transient empty result was observed on 2026-09-20 and the cause is not yet explained. Check the availability field before using them.
• Context sent to the model is capped at 40 000 characters, widened once if needed.
• This is a public beta. Treat it as such.

What Nexora is not

It is not another LLM API. You do not send it a prompt and get a completion. You send it an objective and a budget, and it decides which capabilities to use, runs them in order, checks the result, and hands it back with the evidence and the cost.