AI UGC

Call the RawUGC API to generate AI videos/images/music, manage content (personas, products, styles, characters), schedule social media posts, research TikTok...

Security Scan

The skill is internally consistent with its description: it's an instruction-only adapter for the RawUGC API and only requests a single RawUGC API key needed to call those endpoints.

Purpose & Capability

Name/description (generate videos/images/music, manage personas/products/styles/characters, schedule posts, research TikTok) match the documented API endpoints in SKILL.md and reference.md. The only required env var (RAWUGC_API_KEY) is exactly what an API client needs.

Instruction Scope

SKILL.md gives concrete, scoped instructions to read RAWUGC_API_KEY and call https://rawugc.com/api/v1 endpoints with an appropriate version header. It does not instruct reading unrelated files, other credentials, or exfiltrating data to third-party endpoints. It warns not to log the key.

Install Mechanism

No install spec and no code files (instruction-only). Nothing is written to disk or downloaded during install, which minimizes surface area.

Credentials

Only RAWUGC_API_KEY is required. No unrelated secrets, config paths, or multiple credentials are requested. The credential requested is proportional to the skill's purpose.

Persistence & Privilege

The skill is not forced-always, does not request system-wide changes, and contains no install-time persistence. disable-model-invocation is false (agent may call the skill autonomously), which is the platform default and not itself a red flag here.

Assessment

This skill appears coherent and only needs your RawUGC API key. Before installing, confirm you trust rawugc.com and that the API key you provide has minimal necessary scope; be aware uploaded or generated media could be published if your RawUGC account is linked to social accounts (check RawUGC's dashboard and permissions). Because the skill can be invoked by the agent, consider whether you want the agent to call RawUGC autonomously or prefer to only call it manually. Avoid sharing the API key elsewhere, rotate it if you suspect misuse, and do not provide other unrelated credentials.

RawUGC API

Procedural knowledge for agents to call the RawUGC API. All requests require an API key from the RawUGC dashboard, passed via environment variable.

Authentication

• Environment variable: Read the API key from RAWUGC_API_KEY. The key is created in the RawUGC dashboard and must be kept secret; do not hardcode or log it.
• Header: Send on every request: Authorization: Bearer <value of RAWUGC_API_KEY>.
• If RAWUGC_API_KEY is missing or empty, inform the user they must set it and obtain a key from the RawUGC dashboard.

Base URL

• Production: https://rawugc.com/api/v1
• All paths below are relative to this base.

API Versioning

RawUGC uses date-based API versioning. The current latest version is 2026-03-06.

• RawUGC-Version request header: Override the version per-request (recommended).
• API key pinned version: Set when creating the key in the dashboard.
• Fallback: Latest version (2026-03-06) if neither is set.

Always send RawUGC-Version: 2026-03-06 in requests to ensure consistent behavior.

---

Video Generation

POST /videos/generate

Initiate video generation.

Request body (JSON):

Field	Type	Required	Description
model	string	Yes	sora-2-text-to-video, sora-2-image-to-video, kling-2.6/motion-control, veo3, veo3_fast
prompt	string	For text-to-video / veo3	Text description (1-5000 chars)
imageUrls	string[]	For image-to-video / kling	URLs, max 10. Veo3/veo3_fast accept up to 2 optional images.
videoUrls	string[]	For kling	URLs, max 1. Required for kling-2.6/motion-control
aspectRatio	string	No	Sora: portrait/landscape. Veo3: 16:9/9:16/Auto
nFrames	string	No	"10" or "15" (Sora only)
selectedCharacter	string	No	Character username (e.g. rawugc.mia)
characterOrientation	string	No	image or video (kling only)
mode	string	No	720p or 1080p (kling only)

Response (201): videoId, model, status, creditsUsed, newBalance, estimatedCompletionTime, createdAt.

GET /videos/:videoId

Get video status. Returns videoId, status, model, prompt, creditsUsed, url (when completed), createdAt, completedAt, failCode, failMessage, versions (edit history array).

GET /videos

List videos. Query: status, limit (1-100, default 50), page. Returns videos array + pagination.

POST /videos/captions

Add styled captions to a completed video. Costs 1 credit.

Field	Type	Required	Description
videoId	string	Yes	Video identifier (vid_xxx)
language	string	No	Language code (e.g. en). Defaults to auto-detect

Response (200): videoId, url, version, operation, creditsUsed.

POST /videos/overlay

Add text overlay to a completed video.

Field	Type	Required	Description
videoId	string	Yes	Video identifier (vid_xxx)
text	string	Yes	Overlay text (1-500 chars)
position	string	No	top, center, or bottom
fontSize	integer	No	8-200 pixels
topBottomMargin	integer	No	0-500 pixels
strokeThickness	number	No	0-10

Response (200): videoId, url, version, operation, creditsUsed.

---

Image Generation

POST /images/generate

Generate AI images using Nano Banana models. Async -- poll GET /images/:imageId.

Field	Type	Required	Description
model	string	Yes	nano-banana-2 (text-to-image, 4 credits) or google/nano-banana-edit (image editing, 2 credits)
prompt	string	Yes	Text description or edit instruction (1-20000 chars)
imageUrls	string[]	For editing	Source images. Required for google/nano-banana-edit. Optional for nano-banana-2 (reference images, max 14).
aspectRatio	string	No	For nano-banana-2: 1:1, 16:9, 9:16, auto, etc.
imageSize	string	No	For google/nano-banana-edit: 1:1, 16:9, 9:16, auto, etc.
resolution	string	No	For nano-banana-2: 1K, 2K, 4K
outputFormat	string	No	png, jpeg, jpg
googleSearch	boolean	No	Use Google Web Search grounding (nano-banana-2 only)

Response (201): imageId, model, status, creditsUsed, newBalance, estimatedCompletionTime, createdAt.

GET /images/:imageId

Get image status. Returns imageId, status, model, prompt, url (when completed), imageSize, resolution, outputFormat, creditsUsed, createdAt, completedAt, failCode, failMessage.

GET /images

List images. Query: status, limit (1-100, default 20), page. Returns images array + pagination.

---

Music Generation

POST /music/generate

Generate AI music using Suno models. 3 credits per generation. Async -- poll GET /music/:musicId.

Field	Type	Required	Description
prompt	string	Yes	Music description (1-2000 chars)
model	string	No	V3_5, V4, V4_5, V4_5PLUS, V4_5ALL, V5 (default: V5)
instrumental	boolean	No	Instrumental only, no vocals (default: true)
title	string	No	Track title (max 200 chars). Enables custom mode with style.
style	string	No	Style descriptor (max 500 chars, e.g. lo-fi hip hop)

Response (201): musicId, model, status, creditsUsed, newBalance, estimatedCompletionTime, createdAt.

GET /music/:musicId

Get music status. Returns musicId, status, model, prompt, audioUrl (when completed), albumArtUrl, duration, title, creditsUsed, createdAt, completedAt, failCode, failMessage.

GET /music

List music tracks. Query: status, limit (1-100, default 20), page. Returns tracks array + pagination.

---

Upload

POST /upload

Upload a video or image file. Returns a URL for use in generation requests (imageUrls, videoUrls) or analyze-video. Max 100MB.

Request: multipart/form-data with file field. Accepted types: video/mp4, video/quicktime, video/webm, image/png, image/jpeg, image/webp.

Response (200): url, contentType, size.

---

Characters

GET /characters

List all available AI characters (built-in + custom). Returns characters array, count, adminCount, userCount.

GET /characters/:characterId

Get a character by ID. Returns _id, username, displayName, description, videoPreviewUrl, type (admin/user), isActive, createdAt, updatedAt.

---

Personas (CRUD)

Personas define target audiences for content plan generation.

• GET /personas -- List all. Returns personas array + count.
• POST /personas -- Create. Body: name (required, max 200), description (required, max 5000). Returns id.
• GET /personas/:personaId -- Get one.
• PATCH /personas/:personaId -- Update. Body: name, description (both optional).
• DELETE /personas/:personaId -- Delete.

PersonaResponse: _id, organizationId, name, description, createdAt, updatedAt.

---

Messaging (CRUD)

Brand/positioning messaging templates.

• GET /messaging -- List all. Returns messages array + count.
• POST /messaging -- Create. Body: name (required, max 200), body (required, max 5000). Returns id.
• GET /messaging/:messageId -- Get one.
• PATCH /messaging/:messageId -- Update. Body: name, body (both optional).
• DELETE /messaging/:messageId -- Delete.

MessagingResponse: _id, organizationId, name, body, createdAt, updatedAt.

---

Products (CRUD)

Products for video generation.

• GET /products -- List all. Returns products array + count.
• POST /products -- Create. Body: name (required, max 200), photos (required, URL array), description (max 1000), messaging (max 5000). Returns id.
• GET /products/:productId -- Get one.
• PATCH /products/:productId -- Update. Body: name, description, photos, messaging (all optional).
• DELETE /products/:productId -- Delete.

ProductResponse: _id, name, description, photos, messaging, createdAt, updatedAt.

---

Styles (CRUD)

Video/image creative styles with optional prompt templates.

• GET /styles -- List all (built-in + custom). Query: type (video/image). Returns styles array + count.
• POST /styles -- Create. Body: name (required, max 200), description (max 1000), type (video/image), aspectRatio (portrait/landscape/square), promptTemplate (max 5000, supports {productName}, {messaging}, {character} placeholders). Returns id.
• GET /styles/:styleId -- Get one.
• PATCH /styles/:styleId -- Update. All fields optional.
• DELETE /styles/:styleId -- Delete.

StyleResponse: _id, name, description, type, aspectRatio, styleId, promptTemplate, isAdmin, isStandard.

---

Social Scheduling

GET /social/accounts

List connected social accounts (max 3 per org). Returns accounts array + count. Each account: accountId, platform (tiktok/instagram/youtube), username, displayName, profilePicture, isActive.

POST /social/accounts

Sync connected accounts from the scheduling provider. Returns { success: boolean }.

DELETE /social/accounts/:accountId

Disconnect a social account. Returns { success: boolean }.

POST /social/posts

Schedule, draft, or immediately publish a video to social media.

Field	Type	Required	Description
videoUrl	string	Yes	URL of video to post
accountIds	string[]	Yes	Target account IDs
mode	string	Yes	schedule, draft, or now
scheduledFor	integer	For schedule	Unix timestamp (ms)
timezone	string	No	IANA timezone (default: UTC)
content	string	No	Caption (max 2200 chars)
videoId	string	No	RawUGC video ID to link
publishToInbox	boolean	No	Send to TikTok Creator Inbox
tiktokPrivacyLevel	string	No	SELF_ONLY, PUBLIC_TO_EVERYONE, MUTUAL_FOLLOW_FRIENDS, FOLLOWER_OF_CREATOR
tiktokAllowComment	boolean	No	Allow TikTok comments
tiktokAllowDuet	boolean	No	Allow TikTok duets
tiktokAllowStitch	boolean	No	Allow TikTok stitches
tiktokCommercialContentType	string	No	none, brand_organic, brand_content

Response (201): SocialPost object.

GET /social/posts

List posts. Query: fromDate (ms), toDate (ms), includeDrafts (boolean). Returns posts array + count.

GET /social/posts/:postId

Get a post.

PATCH /social/posts/:postId

Update a post. Body: content, scheduledFor, timezone, accountIds (at least one field required).

DELETE /social/posts/:postId

Delete a post. Returns { success: boolean }.

POST /social/posts/:postId/reschedule

Reschedule a post. Body: scheduledFor (required, ms), timezone.

POST /social/posts/:postId/publish

Immediately publish a draft post.

SocialPost: postId, platforms, status (draft/scheduled/published/failed), scheduledFor, timezone, content, videoUrl, createdAt, publishedAt.

---

Viral Library

GET /viral-library/videos/:videoId

Get a viral library video with full AI analysis (hooks, keyframes, performance insights). Returns ViralLibraryVideo.

GET /viral-library/search

Semantic search across analyzed videos. Query: q (required, natural language), limit (1-50, default 20). Returns results (array of { video, score }), query, total.

---

Research

POST /scrape-tiktok

Scrape TikTok videos. Costs 3 credits.

Field	Type	Required	Description
query	string	Yes	Search keyword, hashtag, or query (max 500)
mode	string	No	keyword, hashtag, search (default: keyword)
limit	integer	No	1-10 (default: 10)

Response (200): scrapeId (use with content-plans), count, videos (array with id, url, author, description, stats, duration, hashtags, thumbnail, videoUrl).

POST /content-plans

Generate a content plan from scraped videos. Costs 3 credits.

Field	Type	Required	Description
scrapeId	string	Yes	From scrape-tiktok response
brief	string	Yes	Content plan goals (max 5000)

Response (200): planId, scrapeId, brief, topWins, gapsToTest, blueprints (array with category, strategy, evidence, contentIdeas).

GET /content-plans

List all content plans. Returns plans array + count.

POST /analyze-video

Analyze any video URL (social links or direct URLs). Costs 1 credit. Max 150MB.

Field	Type	Required	Description
videoUrl	string	Yes	Video URL to analyze
prompt	string	No	Custom analysis prompt (max 5000)

Response (200): summary, hook, keyframes (array with timestamp, type, description, visual, audio, text), durationSeconds, tags, whyItPerformed, attributesToCopy, hooksToTest.

---

Errors

All error responses use RFC 7807 Problem Details (JSON): type, title, status, detail, instance, errors.

Status	Meaning
400	Validation error. Surface detail and errors to user.
401	Auth error. Check RAWUGC_API_KEY.
402	Insufficient credits. Add credits in dashboard.
403	Insufficient scope. API key lacks permissions.
404	Resource not found.
429	Rate limit exceeded. Check X-RateLimit-Reset header.
500	Server error. Retry or contact support.

Rate Limits

• API Key: 10 req/min. Session: 20 req/min.
• Headers: X-RateLimit-Limit, X-RateLimit-Remaining, X-RateLimit-Reset (unix timestamp).

Workflow: Generate then poll

1. Generate: POST to the generation endpoint. Note the returned ID (videoId/imageId/musicId).
2. Poll: GET the status endpoint periodically (10-30s). Use exponential backoff.
3. Finish: When status === 'completed', use the result URL. When failed, surface error to user.
4. Edit (video only): POST to /videos/captions or /videos/overlay.

For full request/response shapes, see reference.md.