# ZebraTruth — Compliance Checks for Video

> ZebraTruth is a compliance-intelligence API for AI-generated video ads. Upload a video, get back timestamped findings against FTC, EU UCPD, UK ASA, platform-specific rules (TikTok, YouTube, Meta), brand detection, and rights clearance. Built for ad agencies and creators running automated pre-publish gates.

## Primary endpoint — Video Compliance

```
POST https://api.zebratruth.ai/v1/compliance/check-video
```

The video flow is **async** (3 steps + poll):

1. `POST /v1/compliance/media/upload-url` — get a signed Azure Blob upload URL
2. `PUT <uploadUrl>` — upload video bytes (max 200 MB)
3. `POST /v1/compliance/check-video` — submit `{requestId, blobPath, jurisdictions, platforms, mode}`, returns 202
4. `GET /v1/compliance/check-video/{requestId}` — poll until `status: completed`

Reports cite stable evidence IDs: `audio_NNN` (transcript segments), `frame_NNN` (sampled frames), `ocr_NNN` (on-screen text), `signal_NNN` (visual signals — brands, watermarks, restricted categories).

Per-second pricing. 3-180 second videos. mp4 / webm / quicktime.

## Video endpoint family

- `POST /v1/compliance/media/upload-url` — request a SAS upload URL (step 1)
- `POST /v1/compliance/check-video` — submit for analysis (step 3)
- `GET /v1/compliance/check-video/{requestId}` — poll for status + report (step 4)
- `GET /v1/compliance/webhooks/video/secret` — get/create the tenant webhook signing secret
- `POST /v1/compliance/webhooks/video/secret/rotate` — rotate the webhook secret

## Authentication

Bearer token. Get a key at [developers.zebratruth.ai](https://developers.zebratruth.ai):

```
Authorization: Bearer zt_live_<your_key>
```

Test keys (`zt_test_*`) return deterministic sandbox responses with no credit cost.

## Documentation

- Quickstart (video): https://docs.zebratruth.ai/quickstart
- Full guide: https://docs.zebratruth.ai/guides/video-compliance-check
- Concept page (architecture + schema): https://docs.zebratruth.ai/video
- API reference (OpenAPI): https://docs.zebratruth.ai/api-reference/overview

## LLM Skill (for autonomous agents)

ZebraTruth ships a Claude Code / OpenClaw plugin so autonomous agents can run video compliance checks without per-call docs lookup:

```
/plugin marketplace add dafstar76/zebratruth-ai-labs
/plugin install zebratruth-compliance@zebratruth-ai-labs
```

Workflow doc the skill loads on demand:
https://raw.githubusercontent.com/dafstar76/zebratruth-ai-labs/master/skills/zebratruth-compliance/workflows/video-compliance-check.md

## Other content types

ZebraTruth also supports compliance on single images (`POST /v1/compliance/check-image`, sync, multipart) and text/ad copy (`POST /v1/compliance/check`, sync, JSON). Most customers — ad agencies and creators — use the video endpoint. See https://docs.zebratruth.ai/concepts/other-content-types for image + text references.

## Status state machine (video)

```
queued → processing → completed (reportStatus: complete | insufficient_evidence)
                    ↓
                    failed (reportStatus: analysis_failed)
```

Report fields:
- `report.checks[]` — VideoComplianceCheck array with `evidenceIds` + `timelineLocation`
- `report.analysisDiagnostics[]` — pipeline/processing issues (not compliance findings)
- `report.preprocessing` — transcriptionStatus, frameAnalysisStatus, degraded, evidenceTruncated

## Repository

- Docs source: https://github.com/dafstar76/zebratruth-ai-labs
- Dashboard / API key issuance: https://developers.zebratruth.ai
- Status page: https://status.zebratruth.ai