> ## Documentation Index
> Fetch the complete documentation index at: https://docs.vidjutsu.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# vidjutsu compliance

> Scan a video or caption against platform TOS and return a risk score with cited policy clauses.

Flags a video or text against the target platform's current community guidelines or ad standards. Returns a deterministic risk score (0–100), a verdict, and each violation with the verbatim policy clause + source URL it's based on.

**Informational only, not legal advice** — every response carries a disclaimer. See [Terms of Service](/legal/terms#3-not-legal-advice) for the full posture.

## Subcommands

| Command                      | Use when                                                                                             |
| ---------------------------- | ---------------------------------------------------------------------------------------------------- |
| `vidjutsu compliance video`  | You have a video URL and want both visual + transcript analysis                                      |
| `vidjutsu compliance prompt` | You have text (caption, ad copy, draft script, description) you want to pre-flight before publishing |

## `vidjutsu compliance video`

```bash theme={null}
vidjutsu compliance video --videoUrl <url> --platform <platform>
```

| Flag         | Required | Description                                                    |
| ------------ | -------- | -------------------------------------------------------------- |
| `--videoUrl` | Yes      | Public URL of the video to scan                                |
| `--platform` | Yes      | `youtube` \| `tiktok` \| `instagram` \| `meta-ads` \| `ftc`    |
| `--format`   | No       | `shortform` \| `longform`. Inferred from duration when omitted |

**Rate limit:** 20 per day.

## `vidjutsu compliance prompt`

```bash theme={null}
vidjutsu compliance prompt --text "..." --platform <platform>
```

| Flag         | Required | Description                                                                                         |
| ------------ | -------- | --------------------------------------------------------------------------------------------------- |
| `--text`     | Yes      | Caption, ad copy, script, description, etc.                                                         |
| `--platform` | Yes      | `youtube` \| `tiktok` \| `instagram` \| `meta-ads` \| `ftc`                                         |
| `--kind`     | No       | `caption` \| `script` \| `ad-copy` \| `description` \| `other` (label only; doesn't affect scoring) |

**Rate limit:** 100 per day.

## Example

```bash theme={null}
vidjutsu compliance prompt \
  --text "Get rich in 30 days guaranteed — insider investment tip" \
  --platform instagram \
  --kind caption

{
  "riskScore": 45,
  "verdict": "high-risk",
  "violations": [
    {
      "ruleId": "instagram-misinformation-financial-claims",
      "category": "misinformation",
      "severity": "high",
      "explanation": "The caption promises guaranteed returns in a short timeframe, which falls under Meta's misinformation policy on misleading financial claims.",
      "citation": {
        "text": "...",
        "sourceUrl": "https://transparency.meta.com/policies/community-standards/misinformation",
        "policyVersion": "2026-04-23"
      },
      "evidence": {
        "transcriptExcerpt": "Get rich in 30 days guaranteed"
      }
    }
  ],
  "platform": "instagram",
  "policySnapshotDate": "2026-04-23",
  "disclaimer": "This response is informational only and does not constitute legal advice..."
}
```

## How it works

Our ingestion pipeline weekly scrapes each platform's policy pages (YouTube Community Guidelines, TikTok ads + CG, Meta Community Standards + Ad Standards) and extracts the enforceable clauses. On a scan:

1. We analyze the input and retrieve the most relevant policy clauses using vector search
2. An LLM cross-references the content against those clauses
3. Every citation in the response is drawn verbatim from our ingested corpus — no fabricated rules
4. The risk score is deterministic: severity weights sum to a 0–100 value, capped at 100

## Rate limits

* `video` — 20 calls per day
* `prompt` — 100 calls per day

Separate buckets. Check remaining capacity with `vidjutsu usage`.