# Astra Human Preference Oracle – Review Submission

> Astra Human Preference Oracle – Review Submission is a paid API for AI agents from astra-human-preference-oracle.neik420.chatgpt.site, paid per call via x402, $0.01/call, status unknown (last checked 2026-09-14).

Submits a previously created review request for genuine human judgment on preference and naturalness, returning a polling URL to retrieve the result.

## Facts

- Endpoint: POST https://astra-human-preference-oracle.neik420.chatgpt.site/api/review
- Price: $0.01/call
- Payment: x402
- Status: unknown
- Last checked: 2026-09-14
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/astra-human-preference-oracle-review-submission-df8671a2
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_NhlSNhf-Yx6N72sqWo5xl

Status and success rate cover calls made through Zero and Zero's own probes. Third-party monitors may report differently.

## How to call it through Zero

Zero handles the 402 payment challenge and records the run. With the Zero CLI installed (`npm i -g @zeroxyz/cli`):

```sh
zero fetch --capability astra-human-preference-oracle-review-submission-df8671a2 -d '<json body>'
```

Example prompt: Submit request 550e8400-e29b-41d4-a716-446655440000 to the Astra Human Preference Oracle for a real human to judge which response sounds more natural, and give me the poll URL so I can check back for the result.

## When to prefer this

Choose this endpoint when you need genuine human judgment — not AI scoring — on text naturalness or pairwise preference, especially for collecting RLHF training labels, validating AI outputs before deployment, or when regulatory or quality requirements demand human-in-the-loop evaluation. Prefer it over automated LLM-as-judge approaches when human authenticity is required at low per-call cost ($0.01 USDC).

## Known failure modes

- Invalid or expired request_id UUID returns an error or 404
- Missing or invalid Bearer access_token returns 401 Unauthorized
- Payment failure via x402 prevents the call from proceeding
- Request already reviewed or cancelled may return a terminal status rather than pending
- Network or service downtime on chatgpt.site hosting may cause 5xx errors

## How this service works

Buy genuine human preference and naturalness judgment with x402. 0.01 USDC on Base.

## Output

Returns a JSON object with a status field (e.g. 'pending_human_review'), the original request_id, and a poll_url pointing to where the agent can later retrieve the human reviewer's preference judgment once it is complete.

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "properties": {
  "request_id": {
   "type": "string",
   "format": "uuid",
   "description": "ID returned by POST /api/requests. Send its access_token as Authorization: Bearer."
  }
 }
}
```

## Response schema (JSON Schema)

```json
{
 "type": "json",
 "example": {
  "status": "pending_human_review",
  "poll_url": "https://astra-human-preference-oracle.clever-cream-3122.chatgpt.site/api/requests/00000000-0000-4000-8000-000000000001",
  "request_id": "00000000-0000-4000-8000-000000000001"
 }
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/astra-human-preference-oracle-review-submission-df8671a2/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from astra-human-preference-oracle.neik420.chatgpt.site](https://www.zero.xyz/host/astra-human-preference-oracle.neik420.chatgpt.site/llms.txt)