视频反向提示

Analyze videos to extract reverse prompts, shot-by-shot breakdowns, and AI-ready visual descriptions via the NanoPhoto.AI Video Reverse Prompt API. Use when:...

Security Scan

The skill is internally consistent with its stated purpose: it needs only a NanoPhoto API key and the included instructions and script simply upload MP4s (<=30MB) to nanophoto.ai for reverse-prompt analysis; no unrelated credentials, suspicious endpoints, or hidden exfiltration were found.

Purpose & Capability

Name, description, and runtime artifacts align: the skill calls https://nanophoto.ai/api/sora-2/reverse-prompt and requires a single NANOPHOTO_API_KEY which is appropriate for this API-based video analysis purpose. Note: repository metadata (_meta.json) and registry metadata show minor mismatches (ownerId, slug, and version differences) which affect provenance but do not change runtime behavior.

Instruction Scope

SKILL.md instructions are scoped to collecting a video source (YouTube/URL/local .mp4), optionally base64-encoding a local file, and POSTing it to the NanoPhoto API. The instructions explicitly require user authorization for uploads and do not instruct the agent to read unrelated files or environment variables.

Install Mechanism

No install spec (instruction-only) and a small included Python uploader script. The script uses stdlib urllib, validates file extension/size, base64-encodes, and POSTs to the documented endpoint — low install risk.

Credentials

Only NANOPHOTO_API_KEY is required and declared as the primary credential. That is proportionate to the skill's function; no other SECRET/TOKEN/PASSWORD env vars are requested.

Persistence & Privilege

Skill does not request always:true, does not alter other skills' configs, and is not granted elevated persistence. Default autonomous invocation remains (platform normal behavior) but is not combined with other red flags.

Assessment

This skill appears coherent and limited to uploading .mp4 files to NanoPhoto.AI using your NANOPHOTO_API_KEY. Before installing: (1) Provide the API key only via the platform's secure env-variable setting (do not paste into chat). (2) Only upload videos you are authorized to share — uploaded content will leave your environment and go to nanophoto.ai. (3) Check the service's pricing/credits and rate limits. (4) Note minor metadata/provenance mismatches in the package (ownerId/slug/version differences); if provenance matters to you, verify the publisher on https://nanophoto.ai and confirm this package was published by the official NanoPhoto vendor. (5) Use the bundled script for local uploads as recommended (it enforces the 30 MB limit and avoids large inline base64 in shells).

Video Reverse Prompt

Analyze videos to extract detailed shot breakdowns and AI-ready prompts via the NanoPhoto.AI API.

Use the bundled script for local file uploads instead of inlining large base64 payloads directly in the shell; it is more reliable for multi-megabyte videos.

Prerequisites

1. Obtain an API key at: https://nanophoto.ai/settings/apikeys
2. Configure NANOPHOTO_API_KEY before using the skill.

Preferred OpenClaw setup:

• Open the skill settings for this skill
• Add an environment variable named NANOPHOTO_API_KEY
• Paste the API key as its value

Equivalent config shape:

{
  "skills": {
    "entries": {
      "video-reverse-prompt": {
        "enabled": true,
        "env": {
          "NANOPHOTO_API_KEY": "your_api_key_here"
        }
      }
    }
  }
}

Other valid ways to provide the key:

• Shell: export NANOPHOTO_API_KEY="your_api_key_here"
• Tool-specific env config: any runtime that injects NANOPHOTO_API_KEY

Credential declaration summary:

• Required env var: NANOPHOTO_API_KEY
• Primary credential: NANOPHOTO_API_KEY
• No unrelated credentials are required

If the env var is not set, ask the user to configure it before proceeding.

Workflow

1. Collect the video source from the user (YouTube link, direct .mp4 URL, or local file path)
2. Determine videoSource type: youtube, url, or file
3. Confirm the user is authorized to upload or process the content
4. If local file: read and base64-encode it (must be .mp4, max 30 MB)
5. Call the API (streaming response)
6. Return the shot breakdown and prompts to the user

API Call

YouTube Video

curl -X POST "https://nanophoto.ai/api/sora-2/reverse-prompt" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $NANOPHOTO_API_KEY" \
  --data-raw '{
    "videoSource": "youtube",
    "locale": "en",
    "videoUrl": "https://www.youtube.com/watch?v=XXXXXXXXXXX"
  }'

Direct Video URL

curl -X POST "https://nanophoto.ai/api/sora-2/reverse-prompt" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $NANOPHOTO_API_KEY" \
  --data-raw '{
    "videoSource": "url",
    "locale": "en",
    "videoUrl": "https://example.com/video.mp4"
  }'

Local File (Preferred: bundled script)

python3 scripts/reverse_prompt_file.py your-video.mp4 --locale en

The script reads NANOPHOTO_API_KEY from the environment, validates the file size/format, base64-encodes the MP4, and prints the streaming text response.

Local File (Manual Base64 request)

VIDEO_BASE64=$(base64 < your-video.mp4)

curl -X POST "https://nanophoto.ai/api/sora-2/reverse-prompt" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $NANOPHOTO_API_KEY" \
  --data-raw "{
    \"videoSource\": \"file\",
    \"locale\": \"en\",
    \"videoFile\": \"$VIDEO_BASE64\",
    \"videoFileName\": \"your-video.mp4\"
  }"

Parameters

Parameter	Type	Required	Description
videoSource	string	Yes	youtube, url, or file
locale	string	No	Output language (default: en). Supported: en, zh, zh-TW, ja, ko, es, fr, de, pt, ru, ar
videoUrl	string	Conditional	YouTube link or direct .mp4 URL
videoFile	string	Conditional	Base64-encoded video (when videoSource is file)
videoFileName	string	No	Original filename for uploaded videos

Constraints

• Only .mp4 format supported
• Max file size: 30 MB (before base64 encoding)
• videoFile accepts plain base64 or Data URL (data:video/mp4;base64,...)
• Costs 1 credit per API call

Response

The API returns a streaming text response (Content-Type: text/plain; charset=utf-8) containing a Markdown table with:

• Shot number, framing/angle, camera movement
• Detailed visual description
• Audio analysis (BGM, sound effects, narration)
• Duration per shot
• Overall summary

Error Handling

errorCode	HTTP	Cause	Action
LOGIN_REQUIRED	401	Invalid or missing API key	Verify key at https://nanophoto.ai/settings/apikeys
API_KEY_RATE_LIMIT_EXCEEDED	429	Rate limit exceeded	Wait and retry
INSUFFICIENT_CREDITS	402	Not enough credits	Top up credits
INVALID_INPUT	400	Missing required parameters	Check videoSource and corresponding fields
INVALID_YOUTUBE_URL	400	Invalid YouTube URL	Ask user for a valid YouTube link
INVALID_VIDEO_URL	400	Invalid video URL	Ask user for a valid .mp4 URL
INVALID_FORMAT	400	Not MP4 format	Only .mp4 is supported
FILE_TOO_LARGE	400	File exceeds 30 MB	Ask user for a smaller file
VIDEO_DOWNLOAD_FAILED	400	Cannot download video	Check URL accessibility
VIDEO_PROCESSING_FAILED	422	Processing error	Retry or try a different video
AI_SERVICE_ERROR	503	AI service unavailable	Retry later

Bundled script

• scripts/reverse_prompt_file.py: Reliable local-file uploader for .mp4 inputs. Use it when the user provides a local video path.

Full API Reference

See references/api.md for complete endpoint documentation.