# Vevang AI Output Verifier

> Vevang AI Output Verifier is a paid API for AI agents from verifier.vevang.com, paid per call via x402, $1/call, status unknown (last checked 2026-10-01).

Independently verifies whether an AI-generated deliverable (URL or inline content) satisfies a given spec before payment is released in machine-to-machine commerce.

## Facts

- Endpoint: POST https://verifier.vevang.com/verify?utm_source=zero.xyz
- Price: $1/call
- Payment: x402
- Status: unknown
- Last checked: 2026-10-01
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/vevang-ai-output-verifier-677e0868
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_5oYd-l-wArcoQLHblWrlK

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 vevang-ai-output-verifier-677e0868 -d '<json body>'
```

Example prompt: Before we release payment, verify that this deliverable at https://cdn.example.com/output.mp4 actually meets the spec: 'a 60-second explainer video about renewable energy with voiceover, delivered in MP4 format' — give me a pass/fail score and list any issues.

## When to prefer this

Choose this endpoint when an AI agent needs an independent, third-party verification of another agent's output before releasing payment or proceeding in a pipeline — especially in x402 machine-to-machine commerce flows. Prefer it over manual review when the deliverable is a URL-hosted file (video, audio, document) or inline text that must be checked for reachability, media integrity, and spec compliance in a single automated step. It is the right choice when auditability and a structured pass/fail score matter more than subjective human review.

## Known failure modes

- Deliverable URL is unreachable or returns a non-200 response — check will fail with reachability error
- Invalid or unsupported media type at the given URL — media validity check fails
- Spec string is missing or too vague for meaningful judgment — score may be unreliable
- Network timeout fetching the deliverable — returns error or failed check
- Inline content too large for the request body — request may be rejected
- Ambiguous deliverable kind causes misclassification — use explicit `kind` enum to avoid

## How this service works

AI Output Verifier — independently verify another agent's deliverable meets its spec BEFORE you pay for it. POST JSON: deliverable (URL or content) + spec. Checks integrity (reachable, valid media, duration, audio) and judges spec-satisfaction. Returns {passed, score, checks, issues}. The trust layer for machine-to-machine commerce.

## Output

Returns a JSON object containing: `passed` (boolean indicating overall pass/fail), `score` (numeric satisfaction score), `checks` (list of individual check results covering reachability, media validity, duration, audio, and spec satisfaction), and `issues` (list of specific problems found). Gives an agent a structured, auditable verdict on whether a deliverable is acceptable.

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "properties": {
  "kind": {
   "enum": [
    "auto",
    "video",
    "url",
    "text"
   ],
   "type": "string",
   "description": "deliverable type (auto-detected if omitted)"
  },
  "spec": {
   "type": "string",
   "description": "What it must satisfy (the brief/requirements)"
  },
  "deliverable": {
   "type": "string",
   "description": "URL or inline content of the deliverable to verify"
  }
 }
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/vevang-ai-output-verifier-677e0868/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from verifier.vevang.com](https://www.zero.xyz/host/verifier.vevang.com/llms.txt)