BytesAgain is a curated directory of 60,000+ AI agent skills from ClawHub, GitHub, LobeHub, and Dify. Search skills by keyword in 7 languages, browse by role (developer, creator, trader, marketer) or by use case.

How do I find AI skills on BytesAgain?

Use the search bar on BytesAgain.com to search by keyword in 7 languages. You can also browse by role (developer, creator, trader, marketer) or by use case. Each skill shows install instructions for Claude, Cursor, OpenClaw, Continue, and more.

Yes, BytesAgain is completely free. No registration required for searching skills. The MCP API is also free with rate limits.

Does BytesAgain have an API for AI agents?

Yes! BytesAgain provides a free MCP SSE endpoint at /api/mcp/sse for AI agents, plus a REST API at /api/mcp?action=search&q= . No authentication needed.

Can I request a new AI skill on BytesAgain?

Yes! Visit the Requests page on BytesAgain.com to submit a skill request. Your request will be visible to the community and notified to the site admin.

🦀 ClawHub

Content Generator - Articles, Social Posts, Videos & Images from Any Topic or URL

by @osaket

All-in-one content generation, social media publishing, video creation, and image template design via the GenerateBot API. Search trending topics, write blog...

Versionv1.0.0

Downloads616

Stars⭐ 1

#crypto-defi #writing #video #social-media #legal

View on ClawHub →

TERMINAL

clawhub install content-generator-from-url

📖 About This Skill

name: generatebot description: All-in-one content generation, social media publishing, video creation, and image template design via the GenerateBot API. Search trending topics, write blog posts and articles, create short-form video reels, design and render image templates, publish to LinkedIn/Instagram/TikTok/WordPress, enrich with SEO links, generate AI images, and rewrite in brand voice. Use when the user mentions GenerateBot, wants to generate content, create videos, design post images, publish to social media or CMS, or automate content workflows. emoji: 🤖 homepage: https://generatebot.com metadata: openclaw: primaryEnv: GENERATEBOT_API_KEY requires: env: - GENERATEBOT_API_KEY

GenerateBot Core - Search, Generate & Manage Content

Base URL: https://generatebot.com/api/v1

Authentication: Bearer token in the Authorization header.

Authorization: Bearer GENERATEBOT_API_KEY

All request and response bodies use JSON. Set Content-Type: application/json.

Quick Reference

| Action | Endpoint | Credits | Type | |--------|----------|---------|------| | Search content | POST /agents/news-aggregator | 10 | sync | | Fetch RSS feeds | POST /rss | 0 | sync | | Manage RSS feeds | GET/PUT/DELETE /rss | 0 | sync | | Generate from URL | POST /pipelines (content-analyzer) | 100 | async | | Generate from topic | POST /pipelines (topic-to-content) | 100 | async | | Generate scripts | POST /pipelines (script-generator) | 15 | async | | Poll status | GET /pipelines/{runId} | 0 | poll | | List runs | GET /pipelines | 0 | sync | | Save content | POST /content | 0 | sync | | List content | GET /content | 0 | sync | | Get/Update/Delete | GET/PATCH/DELETE /content/{id} | 0 | sync | | Check credits | GET /credits | 0 | sync | | Credit transactions | GET /credits/transactions | 0 | sync |

1. Search for Content Ideas

POST /api/v1/agents/news-aggregator (10 credits)

{
  "topic": "sustainable architecture trends",
  "category": "technology",
  "searchConfig": {
    "resultLimit": 10,
    "countryCode": "us",
    "freshness": "pw",
    "searchBreadth": "wide",
    "preferredDomains": ["techcrunch.com", "reuters.com"],
    "excludeDomains": ["reddit.com"]
  }
}

topic (required): What to search for

category (optional): technology, finance, health, realestate, marketing, etc.

searchConfig (optional):

- resultLimit: number of results (default 10) - countryCode: "us", "au", "gb", etc. - freshness: "pd" (past day), "pw" (past week), "pm" (past month), "py" (past year) - searchBreadth: "narrow", "balanced", "wide" - preferredDomains / excludeDomains: arrays of domain strings

Returns: array of articles with url, title, description, source, publishedAt, category

Use the returned URLs with the pipeline endpoint below

Tip: RSS feeds (below) are a free alternative to the news aggregator.

2. RSS Feeds (Free Content Sourcing)

All RSS operations cost 0 credits.

POST /api/v1/rss - Fetch and parse feeds

{
  "feedUrls": ["https://techcrunch.com/feed/", "https://blog.example.com/rss"],
  "maxItemsPerFeed": 10,
  "keywords": ["AI", "marketing"],
  "publishedAfter": "2026-04-01T00:00:00Z"
}

feedUrls (optional): Direct RSS/Atom feed URLs (max 50)

urls (optional): Website URLs to auto-discover feeds (max 20)

useSavedFeeds (optional): If true, includes your saved feed URLs

maxItemsPerFeed: 1-100

keywords: Filter items by title + description match (max 20)

publishedAfter: ISO 8601 date filter

feedFilter / feedExclude: Include/exclude feeds by pattern

dedup: Deduplicate items across runs

skipUnhealthyFeeds: Skip feeds with 5+ consecutive failures

digest: { "maxDigestItems": 20, "groupByTopic": true } for summarized output

Must provide at least one of urls, feedUrls, or useSavedFeeds

Returns: feeds array with items (title, link, description, publishedAt, source, etc.)

GET /api/v1/rss - List saved feed URLs

Returns: { "feedUrls": [...], "count": 3 }

PUT /api/v1/rss - Save feed URLs (requires rss:write scope)

Input: { "feedUrls": ["https://example.com/feed.xml"] } (max 50)

Requires an existing business profile

DELETE /api/v1/rss - Clear all saved feed URLs (requires rss:write scope)

3. Generating Content (Pipelines)

POST /api/v1/pipelines (async)

| Pipeline Type | Credits | Requires | Purpose | |---------------|---------|----------|---------| | content-analyzer | 100 | url | Generate articles, social posts, scripts, carousel from a URL | | full-pipeline | 100 | query or queries[] | Generate content from a topic or question | | script-generator | 15 | query | Generate video scripts only |

From a URL (most common):

{ "pipelineType": "content-analyzer", "url": "https://example.com/article" }

From a topic:

{ "pipelineType": "full-pipeline", "query": "How AI is transforming healthcare" }

Optional fields:

market: "au" or "us"

resultLimit: 1-100

businessProfile: object with company context

carouselTemplate: { "prompt": "...", "username": "..." } (content-analyzer only)

Response: { "pipelineRunId": "uuid", "status": "running", "creditsConsumed": 100, "statusUrl": "/api/v1/pipelines/{runId}" }

Save the pipelineRunId -- you need it later as sourcePipelineRunId when creating videos.

Polling: GET /api/v1/pipelines/{runId} every 10 seconds.

Status values: "pending" -> "running" -> "completed" | "failed"

Typical duration: 2-3 minutes. Keep polling up to 5 minutes.

GET /api/v1/pipelines - List pipeline runs

Query params: ?status=completed&pipelineType=content-analyzer&limit=10&offset=0

4. Completed Pipeline Response Shape

When status is "completed", the response contains all generated content:

{
  "status": "completed",
  "results": {
    "data": {
      "agents": {
        "aiContentGenerator": {
          "generatedContent": {
            "articles": [{ "title": "...", "introduction": "...", "sections": [{ "heading": "...", "content": "..." }], "wordCount": 1200 }],
            "linkedInPosts": [{ "intro": "...", "mainContent": "...", "callToAction": "...", "hashtags": ["#AI"] }],
            "instagramPosts": [{ "caption": "...", "hashtags": ["#AI"], "imageTextOverlay": "..." }]
          }
        },
        "scriptGenerator": {
          "scripts": [{
            "hook": "Breaking news you need to know!",
            "platform": "tiktok",
            "mainContent": [
              { "scene": 1, "voiceOver": "Here is what happened...", "textOverlay": "Key Point 1" }
            ],
            "callToAction": "Follow for more!",
            "hashtags": ["#News"]
          }]
        },
        "imageFinder": {
          "foundImages": [
            { "imageUrl": "https://cdn.example.com/img1.jpg", "altText": "...", "suggestedUse": "hero" },
            { "imageUrl": "https://cdn.example.com/img2.jpg", "altText": "...", "suggestedUse": "supporting" }
          ]
        },
        "carouselGenerator": {
          "carousel": {
            "slides": [
              { "slideNumber": 1, "text": "Key takeaway from the article" },
              { "slideNumber": 2, "text": "Supporting point with context" },
              { "slideNumber": 3, "text": "Final thought and call to action" }
            ]
          }
        }
      }
    },
    "totalImagesFound": 30,
    "totalContentItems": 5
  }
}

Key paths to extract:

Articles: results.data.agents.aiContentGenerator.generatedContent.articles[0]

Scripts: results.data.agents.scriptGenerator.scripts[0]

Images: results.data.agents.imageFinder.foundImages

Carousel slides: results.data.agents.carouselGenerator.carousel.slides

IMPORTANT: Pipeline results are NOT auto-saved. Save via POST /content before publishing.

5. Content Library

POST /api/v1/content - Save content (required before CMS/social publishing)

{
  "contentType": "article",
  "title": "Article Title",
  "contentData": {
    "introduction": "Intro paragraph...",
    "sections": [{ "heading": "Section", "content": "Body text..." }],
    "metaDescription": "Short description"
  },
  "tags": ["ai", "marketing"]
}

contentData: JSON object (max 100KB)

tags (optional): max 20 strings

notes (optional): max 2000 chars

Returns: { "content": { "id": "uuid", ... } } -- save the id as contentId

GET /api/v1/content - List content

Query params: ?contentType=article&status=draft&limit=10&offset=0&tags=ai,marketing

Returns: content list with pagination (no full contentData in list view)

GET /api/v1/content/{id} - Get single item with full contentData

PATCH /api/v1/content/{id} - Update (title, contentData, tags, notes, status)

DELETE /api/v1/content/{id} - Delete

6. Credits

GET /api/v1/credits (0 credits) Returns: { "total": 500, "subscription": 400, "oneTime": 100, "adminGranted": 0, "quota": 1000 }

GET /api/v1/credits/transactions (0 credits) - Credit transaction history

Query params: ?limit=20&offset=0

Returns: list of credit transactions with amount, type, description, and timestamps

Error Handling

| HTTP Status | Meaning | |-------------|---------| | 401 | Invalid or missing API key | | 402 | Insufficient credits | | 429 | Rate limit exceeded -- wait and retry | | 400 | Invalid request body -- check error details | | 404 | Resource not found |

All errors: { "error": { "code": "ERROR_CODE", "message": "..." }, "requestId": "..." }

Other GenerateBot Skills

Video Skill (generatebot-video): Create video reels from pipeline data. Step-by-step tutorial for slides, image modes, captions, TTS.

Publish Skill (generatebot-publish): Post to social media, publish to CMS, enrich articles, brand voice, AI images.

Templates Skill (generatebot-templates): Design and render canvas-based post image templates.

Workflow Skill (generatebot-workflows): End-to-end workflow examples, credit costs, usage patterns.

GenerateBot Video - Create Video Reels

Base URL: https://generatebot.com/api/v1

Authentication: Bearer token in the Authorization header.

Authorization: Bearer GENERATEBOT_API_KEY

All request and response bodies use JSON. Set Content-Type: application/json.

Overview

This skill creates short-form video reels (9:16 portrait, 1080x1920) with AI voiceover from content you have generated.

Cost: 200 credits per video

Endpoint: POST /api/v1/videos

Max concurrent renders: 2 (extra requests return 429)

Typical render time: 60-180 seconds

Prerequisites

Before creating a video, you need completed pipeline results. Run a content pipeline first (see Core skill):

POST /api/v1/pipelines
{ "pipelineType": "content-analyzer", "url": "https://example.com/article" }

Poll until status is "completed", then extract these three things:

1. Carousel slides: results.data.agents.carouselGenerator.carousel.slides

   [{ "slideNumber": 1, "text": "Key takeaway from the article" }, ...]

2. Images: results.data.agents.imageFinder.foundImages

   [{ "imageUrl": "https://...", "altText": "...", "suggestedUse": "hero" }, ...]

3. Hook: results.data.agents.scriptGenerator.scripts[0].hook

   "Breaking news you need to know!"

4. Pipeline Run ID: The pipelineRunId from the pipeline POST response.

Step 1: Build Video Slides

Each carousel slide becomes one video slide. Map them like this:

{
  "text": "",
  "imageUrl": "",
  "highlight": ["key", "words"]
}

Pairing rules:

Use the suggestedUse: "hero" image for slide 1

Use remaining images for subsequent slides

Each slide MUST have a unique imageUrl -- never reuse the same image across slides

text is max 200 characters

The text field is both the on-screen caption AND the voiceover narration. Write it as spoken language:

Numbers in words: "25 million dollars" not "$25M"

Use "A.I." not "AI" (TTS pronounces it better)

Add ellipsis for dramatic pauses: "And then... everything changed."

Step 2: Image Display Modes

The imageMode field controls how images appear in the 9:16 portrait frame:

| Mode | Behavior | Best For | |------|----------|----------| | *(omit)* | Auto-detect: landscape (w/h > 1.2) uses overlay, else background | Most cases -- recommended | | "overlay" | Image shown as 16:9 PiP inset (top 12%) over blurred/darkened version of itself | Landscape images, real estate, group photos | | "background" | Full-screen cover-crop fills the entire frame | Portrait images, close-ups | | "background_with_overlays" | Primary image full-bleed + extraImages as PiP insets | Multiple detail shots over a main image |

Recommendation: Omit imageMode in most cases. Auto-detection handles landscape vs portrait correctly. Only set it explicitly to override.

Step 3: Extra Images and Word-Synced Timing

Add related images as overlays that appear synced to specific words in the narration:

{
  "text": "A stunning property in Brisbane with a resort style pool",
  "imageUrl": "https://cdn.example.com/mansion-exterior.jpg",
  "extraImages": [
    "https://cdn.example.com/mansion-aerial.jpg",
    "https://cdn.example.com/mansion-pool.jpg"
  ],
  "extraImageTimings": [
    { "showAtWordIndex": 2 },
    { "showAtWordIndex": 7 }
  ]
}

How word indexing works: Count words in text starting from 0, split on whitespace. Punctuation stays attached.

For the text above: A(0) stunning(1) property(2) in(3) Brisbane(4) with(5) a(6) resort(7) style(8) pool(9)

The aerial shot appears at word 2 ("property")

The pool shot appears at word 7 ("resort")

Each overlay stays visible until the next one starts or the slide ends

When to use extraImages:

Landscape images that lose detail when cropped to 9:16

Multiple angles of the same subject (exterior + pool + interior)

You CAN reuse the main imageUrl in extraImages for blurred-bg + clear-overlay effect

Limits: Max 5 extra images per slide. If you omit extraImageTimings, overlays distribute evenly (less precise).

Step 4: Highlights

The highlight field is an array of words from text to emphasize visually (rendered in accent color with glow effect):

"highlight": ["25 million", "mansion"]

Pick 1-3 key words or phrases that appear in the text

Max 10 items

If omitted, the API auto-generates 2 highlights (prefers numbers and capitalized words)

Step 5: Per-Slide Style Overrides (Optional)

Each slide can override global styles:

| Field | Type | Range | |-------|------|-------| | fontSize | number | 12-120 | | fontWeight | string | "normal" or "bold" | | textColor | string | hex, e.g. "#FFFFFF" | | backgroundColor | string | hex, e.g. "#000000" | | textAlign | string | "left", "center", "right" | | textPosition | string | "top", "center", "bottom" | | imageOpacity | number | 0-1 | | textWidthPercent | number | 10-100 | | imageScale | number | 0.1-5 (zoom level) | | imagePositionX | number | horizontal offset | | imagePositionY | number | vertical offset |

Advanced text styling (textStyle object, all fields optional):

{
  "textStyle": {
    "fontFamily": "Inter",
    "letterSpacing": 1.2,
    "lineHeight": 1.4,
    "textTransform": "uppercase",
    "shadow": { "enabled": true, "offsetX": 2, "offsetY": 2, "blur": 4, "color": "#000000" },
    "outline": { "enabled": true, "width": 2, "color": "#000000" },
    "background": { "enabled": true, "color": "#000000", "opacity": 0.7, "paddingX": 12, "paddingY": 8, "borderRadius": 8 },
    "glow": { "enabled": true, "blur": 10, "color": "#75F30F", "intensity": 0.8 }
  }
}

textTransform: "none", "uppercase", "lowercase", "capitalize"

shadow: Drop shadow behind text

outline: Stroke around text characters

background: Colored box behind text (like a caption box)

glow: Colored glow effect around text

Step 6: Set the Hook

"hook": ""

Max 200 characters

This is the opening text shown before slides begin

No separate script field needed -- the API auto-builds TTS narration from all slide texts

Step 7: Configure Global Style (Optional)

"style": {
  "accentColor": "#FF5500",
  "captionStyle": "outlined",
  "captionPosition": "center",
  "fontSize": 48,
  "colorGrade": "cinematic",
  "filmGrain": { "enabled": true, "opacity": 0.025, "fps": 8 }
}

| Field | Options | Default | |-------|---------|---------| | accentColor | Hex color | #75F30F | | captionStyle | "default", "outlined", "boxed", "marker" | "outlined" | | captionPosition | "center", "lower-third", "top" | "lower-third" | | fontSize | 24-96 | 52 | | colorGrade | "cinematic", "warm", "cool", "vibrant", "none" | none | | filmGrain | { enabled, opacity (0-0.1), fps (1-30) } | disabled |

Step 8: Configure Voice (Optional)

| Field | Description | Default | |-------|-------------|---------| | voiceId | ElevenLabs voice ID | (default Australian male) | | ttsModel | "eleven_v3", "eleven_flash_v2_5", "eleven_multilingual_v2", "eleven_turbo_v2_5" | "eleven_v3" | | speedFactor | 0.5-3.0, post-render speed multiplier | 1.35 |

Step 9: Assemble and Submit

POST /api/v1/videos (200 credits)

{
  "hook": "$25M mansion hits the market!",
  "slides": [
    {
      "text": "A 25 million dollar mansion in Brisbane is up for grabs.",
      "imageUrl": "https://cdn.example.com/mansion-exterior.jpg",
      "highlight": ["25 million", "mansion"],
      "extraImages": ["https://cdn.example.com/mansion-aerial.jpg"],
      "extraImageTimings": [{ "showAtWordIndex": 4 }]
    },
    {
      "text": "Spanning 1400 square meters with five bedrooms and eight bathrooms.",
      "imageUrl": "https://cdn.example.com/mansion-pool.jpg",
      "highlight": ["1400", "bathrooms"]
    }
  ],
  "sourcePipelineRunId": "",
  "style": {
    "accentColor": "#75F30F",
    "captionStyle": "outlined",
    "captionPosition": "center"
  }
}

Field constraints:

hook: 1-200 chars, required

slides: 1-15 slides, required (4-7 recommended)

slides[].text: 1-200 chars, required

slides[].imageUrl: HTTPS URL, required

slides[].highlight: max 10 items, optional

slides[].extraImages: max 5 URLs, optional

watermark: small text in corner, max 100 chars, optional

watermarkLogoUrl: HTTPS URL to a logo image for the watermark, optional

watermarkPosition: position string (max 50 chars), optional

cta: { "text": "Follow for more!", "url": "example.com" }, optional

sourcePipelineRunId: UUID, strongly recommended -- links video to source content

Response:

{
  "jobId": "uuid",
  "contentId": "uuid",
  "status": "queued",
  "creditsConsumed": 200,
  "statusUrl": "/api/v1/videos/{jobId}"
}

Step 10: Poll for Completion

GET /api/v1/videos/{jobId} every 5 seconds.

Status progression: queued -> generating_audio -> rendering -> uploading -> completed

queued: 10-30s while render worker starts. Normal.

generating_audio: TTS is being generated. Progress, not a stall.

rendering: Video frames being composed. Longest phase.

completed: videoUrl contains the download URL.

failed: error field contains the error message.

Keep polling up to 5 minutes. Do NOT stop on intermediate statuses. Do NOT report failure until status is "failed" or you have polled the maximum duration.

Step 11: Post the Video

The video response includes a contentId. Use it directly with POST /api/v1/social/post (see Publish skill):

{
  "contentId": "",
  "platforms": [{
    "accountId": "",
    "platform": "instagram",
    "contentType": "video",
    "caption": "Check out our latest video!"
  }]
}

Render Time Estimates

| Slides | Approximate Time | |--------|------------------| | 3-5 | 60-90 seconds | | 6-10 | 90-150 seconds | | 11-15 | 150-180 seconds |

Audio generation adds 10-20 seconds. Upload adds 5-10 seconds.

Error Handling

| HTTP Status | Meaning | |-------------|---------| | 402 | Insufficient credits (need 200) | | 429 | Too many concurrent renders (max 2) -- wait and retry | | 400 | Invalid request body -- check error details |

Other GenerateBot Skills

Core Skill (generatebot-core): Search for content, run pipelines, manage content library. Required before creating videos.

Publish Skill (generatebot-publish): Post completed videos to social media, publish articles to CMS.

Templates Skill (generatebot-templates): Design and render canvas-based post image templates.

Workflow Skill (generatebot-workflows): End-to-end workflow examples and patterns.

GenerateBot Publish - Social Media, CMS & Content Enhancement

Base URL: https://generatebot.com/api/v1

Authentication: Bearer token in the Authorization header.

Authorization: Bearer GENERATEBOT_API_KEY

All request and response bodies use JSON. Set Content-Type: application/json.

Quick Reference

| Action | Endpoint | Credits | |--------|----------|---------| | List social accounts | GET /social/accounts | 0 | | Post to social | POST /social/post | 0 | | Social post history | GET /social/history | 0 | | List CMS accounts | GET /cms/accounts | 0 | | Publish to CMS | POST /cms/publish | 0 | | CMS publish history | GET /cms/history | 0 | | Create template | POST /templates | 0 | | List templates | GET /templates | 0 | | Get/Update/Delete | GET/PATCH/DELETE /templates/{id} | 0 | | Render template | POST /templates/{id}/render | 20 | | Enrich with links | POST /enrich/internal-links | 20 | | Enrich with images | POST /enrich/images | 20 | | Brand voice rewrite | POST /agents/brand-voice-rewriter | 5 | | Generate images | POST /agents/ai-image-generator | 10 | | Reddit replies | POST /agents/reddit-replier | 10 | | Business profile | POST /agents/business-profile-generator | 5 |

Prerequisites

Most publish actions require a contentId:

For articles: Save via POST /content first (see Core skill). The response id is your contentId.

For videos: The POST /videos response includes a contentId automatically (see Video skill).

1. Post to Social Media (0 credits)

Step 1: List connected accounts: GET /api/v1/social/accounts Returns: [{ "id": "uuid", "platform": "instagram", "username": "@handle", "isConnected": true }]

Step 2: Post content: POST /api/v1/social/post

{
  "contentId": "uuid-of-saved-content",
  "platforms": [
    {
      "accountId": "uuid-from-social-accounts",
      "platform": "instagram",
      "contentType": "video",
      "caption": "Check out our latest video!"
    },
    {
      "accountId": "uuid-linkedin",
      "platform": "linkedin",
      "contentType": "text",
      "text": "Great insights on AI in healthcare..."
    }
  ]
}

contentId (required): UUID from POST /content or POST /videos

platforms[].accountId (required): from GET /social/accounts

platforms[].platform: "instagram", "tiktok", "threads", "linkedin", "youtube"

platforms[].contentType: "text", "image", "video", "carousel", "multi-image"

Optional: caption (max 5000), text (max 5000), hashtags (max 30), mediaUrls (max 10), title (max 200)

2. Publish to CMS (0 credits)

Step 1: List connected CMS platforms: GET /api/v1/cms/accounts Returns: [{ "id": "uuid", "platform": "wordpress", "siteUrl": "https://myblog.com", "isConnected": true }]

Step 2: Publish: POST /api/v1/cms/publish

{
  "contentId": "uuid-from-post-content",
  "cmsAccountId": "uuid-from-cms-accounts",
  "title": "Article Title",
  "content": "Introduction paragraph...
Section Heading
Section content...",
  "excerpt": "Short description",
  "featuredImageUrl": "https://cdn.example.com/hero.jpg",
  "status": "published"
}

contentId (required): UUID from POST /content

cmsAccountId (required): from GET /cms/accounts

title: 1-500 chars

content: HTML string, 1-200,000 chars

Optional: slug (max 200), excerpt (max 1000), featuredImageUrl, categories (max 10), tags (max 20), status ("draft" | "published"), scheduledAt (ISO 8601)

HTML conversion tip: Wrap introduction in

tags, each section as

`heading`

content

3. Enrich Article with Internal Links (20 credits)

POST /api/v1/enrich/internal-links

{
  "article": {
    "title": "Article title",
    "introduction": "Intro text...",
    "sections": [{ "heading": "Section heading", "content": "Section body..." }]
  },
  "businessProfile": {
    "sitemapUrl": "https://example.com/sitemap.xml",
    "websiteUrl": "https://example.com"
  }
}

Returns: enriched article with relevant internal links inserted into the content.

4. Enrich Article with Images (20 credits)

POST /api/v1/enrich/images

{
  "article": {
    "title": "Article title",
    "sections": [{ "heading": "Section heading", "content": "Section body..." }]
  },
  "images": [
    { "imageUrl": "https://...", "altText": "Description", "suggestedUse": "hero" }
  ]
}

suggestedUse: "hero", "supporting", "thumbnail", or "background"

Returns: article with images matched to appropriate sections

5. Rewrite in Brand Voice (5 credits)

POST /api/v1/agents/brand-voice-rewriter

{
  "articles": [{
    "title": "Article Title",
    "introduction": "Article intro...",
    "sections": [{ "heading": "Section heading", "content": "Section body..." }],
    "metaDescription": "Short description"
  }],
  "businessProfile": {
    "companyName": "Acme Corp",
    "brandVoice": "professional",
    "targetAudience": "enterprise CTOs",
    "industryCategory": "technology"
  }
}

Returns: rewritten articles matching the brand voice.

6. Generate AI Images (10 credits)

POST /api/v1/agents/ai-image-generator

{
  "prompts": ["Modern office workspace with natural lighting, minimalist design"],
  "style": "natural",
  "size": "1024x1024"
}

prompts (required): Array of 1-5 strings (min 10 chars each)

style: "natural" or "vivid" (default: "natural")

size: "1024x1024", "1536x1024", "1024x1536", "auto"

quality: "low", "medium", "high"

useCase: "thumbnail", "instagram", "carousel", "article" (auto-selects best size/quality)

Returns: { "data": { "images": [{ "imageUrl": "https://cdn.generatebot.com/...", "prompt": "..." }] } }

Use the imageUrl values directly as slide images when creating video reels

7. Generate Reddit Replies (10 credits)

POST /api/v1/agents/reddit-replier

{
  "redditPostUrl": "https://www.reddit.com/r/smallbusiness/comments/abc123/best_tools/"
}

redditPostUrl (required): Full URL to a Reddit post

Optional: analysisConfig: { "responseStyle": "professional" | "casual" | "educational" }

Returns: post analysis with reply suggestions and confidence scores

8. Generate Business Profile (5 credits)

POST /api/v1/agents/business-profile-generator

{
  "websiteUrl": "https://example.com",
  "companyName": "Example Corp"
}

websiteUrl (required): Company website to crawl

companyName (required): Company name

Optional: userDescription, industryHint, country (default: "Australia")

Returns: structured profile with name, description, industry, targetAudience, keyProducts, tone

9. Post Templates (0 credits for CRUD, 20 credits to render)

Design and render social media post images using Fabric.js canvas templates.

POST /api/v1/templates - Create a template

{
  "name": "LinkedIn Announcement",
  "description": "Clean announcement card for LinkedIn",
  "category": "linkedin",
  "canvasJson": "{...Fabric.js canvas JSON...}",
  "templateMetadata": {
    "fields": [
      { "key": "headline", "objectName": "headline-text", "label": "Headline", "maxLength": 80, "required": true },
      { "key": "body", "objectName": "body-text", "label": "Body Text", "maxLength": 200 }
    ],
    "imageSlots": [
      { "key": "logo", "objectName": "logo-image", "required": true }
    ]
  }
}

category: "linkedin", "instagram", or "article"

canvasJson: Valid Fabric.js JSON string (max 500KB). Must contain an objects array with a clip workspace object.

templateMetadata.fields: Max 20 text fields. Each maps a key to a canvas objectName.

templateMetadata.imageSlots: Max 10 image slots. Each maps a key to a canvas objectName.

GET /api/v1/templates - List templates

Query params: ?category=linkedin&limit=10&offset=0

Returns: templates with pagination (no canvas JSON in list view)

GET /api/v1/templates/{id} - Get template with full canvas JSON

PATCH /api/v1/templates/{id} - Update (name, description, category, canvasJson, templateMetadata, isActive)

DELETE /api/v1/templates/{id} - Delete template

POST /api/v1/templates/{id}/render (20 credits) - Render template to image

{
  "fields": {
    "headline": "We Just Launched!",
    "body": "Our new AI-powered content platform is live."
  },
  "images": {
    "logo": "https://cdn.example.com/logo.png"
  },
  "outputFormat": "jpeg",
  "quality": 0.9
}

fields: Key-value map matching template field keys

images: Key-value map matching template image slot keys (HTTPS URLs only)

outputFormat: "jpeg" (default) or "png"

quality: 0.1-1.0 (default: 0.9)

Returns: { "creditsConsumed": 20, "imageUrl": "https://cdn...", "width": 1080, "height": 1080 }

10. Social Post History (0 credits)

GET /api/v1/social/history - View past social media posts

Query params: ?platform=instagram&status=published&limit=20&offset=0

platform: "instagram", "tiktok", "threads", "linkedin", "youtube"

status: "pending", "uploading", "processing", "published", "failed", "expired", "cancelled"

Returns: list of posts with platform, contentType, caption, status, platformPostId, publishedAt

11. CMS Publish History (0 credits)

GET /api/v1/cms/history - View past CMS publications

Query params: ?platform=wordpress&status=published&limit=20&offset=0

platform: "wordpress", "ghost", "webflow", "wix", "shopify", "framer", "notion"

status: "pending", "publishing", "published", "failed", "updated", "cancelled"

Returns: list with platform, title, status, platformPostUrl, publishedAt, errorMessage

Error Handling

All errors: { "error": { "code": "ERROR_CODE", "message": "..." }, "requestId": "..." }

Other GenerateBot Skills

Core Skill (generatebot-core): Search for content, run pipelines, manage content library, RSS feeds.

Video Skill (generatebot-video): Create video reels from pipeline data.

Templates Skill (generatebot-templates): Design and render canvas-based post image templates.

Workflow Skill (generatebot-workflows): End-to-end workflow examples and patterns.

GenerateBot Templates - Design & Render Post Images

Base URL: https://generatebot.com/api/v1

Authentication: Bearer token in the Authorization header.

Authorization: Bearer GENERATEBOT_API_KEY

All request and response bodies use JSON. Set Content-Type: application/json.

Quick Reference

| Action | Method | Path | Credits | Type | |--------|--------|------|---------|------| | Create template | POST | /templates | 0 | sync | | List templates | GET | /templates | 0 | sync | | Get template | GET | /templates/{id} | 0 | sync | | Update template | PATCH | /templates/{id} | 0 | sync | | Delete template | DELETE | /templates/{id} | 0 | sync | | Render template | POST | /templates/{id}/render | 20 | sync |

System Overview

Templates are canvas JSON documents with dynamic text and image slots. When rendered, the system substitutes placeholder content with real text and images, auto-fits text to its bounding box, and smart-fits images to cover their target areas. The output is a high-quality image (JPEG or PNG) uploaded to CDN.

Canvas JSON Format

#### Top-Level Structure

{
  "version": "5.3.0",
  "objects": [ ... ],
  "clipPath": {
    "type": "rect",
    "version": "5.3.0",
    "left": 0, "top": 0,
    "width": 1080, "height": 1920,
    "fill": "#0a0a0a",
    "scaleX": 1, "scaleY": 1
  }
}

version: Always "5.3.0"

objects: Ordered array of canvas objects (back-to-front rendering order)

clipPath: Canvas-level clip boundary -- MUST match the clip rect dimensions and fill

#### The Clip Rect (Workspace)

Every template MUST have a clip rect as its first object. This defines the artboard:

{
  "type": "rect",
  "version": "5.3.0",
  "originX": "left", "originY": "top",
  "left": 0, "top": 0,
  "width": 1080, "height": 1920,
  "fill": "#0a0a0a",
  "stroke": null, "strokeWidth": 0,
  "scaleX": 1, "scaleY": 1,
  "opacity": 1, "visible": true,
  "name": "clip",
  "selectable": false, "hasControls": false
}

Common dimensions:

| Format | Width | Height | Use Case | |--------|-------|--------|----------| | Portrait / Story | 1080 | 1920 | Instagram Stories, LinkedIn Stories | | Square | 1080 | 1080 | Instagram Feed, LinkedIn Feed | | Landscape | 1200 | 627 | LinkedIn Articles, Twitter/X |

#### Object Naming Convention

| Pattern | Type | Example | Purpose | |---------|------|---------|---------| | {{key}} | Dynamic | {{title}}, {{primaryImage}} | Replaced during rendering | | Plain name | Static | gradient_overlay_1, accent_line | Fixed decorative elements | | clip | Workspace | clip | Artboard boundary (exactly one) |

#### Image Objects

Image placeholders define where dynamic images appear. The src MUST use a 1x1 transparent pixel -- never an empty string.

{
  "type": "image",
  "version": "5.3.0",
  "originX": "left", "originY": "top",
  "left": 0, "top": 0,
  "width": 1200, "height": 800,
  "scaleX": 0.9, "scaleY": 2.4,
  "opacity": 1, "visible": true,
  "name": "{{primaryImage}}",
  "selectable": true, "hasControls": true,
  "src": "data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAQAAAC1HAwCAAAAC0lEQVR42mNkYAAAAAYAAjCB0C8AAAAASUVORK5CYII=",
  "crossOrigin": "anonymous",
  "filters": []
}

Key rules:

src MUST be the 1x1 transparent pixel base64 shown above. Empty string "" causes rendering failures.

crossOrigin MUST be "anonymous"

filters MUST be [] (empty array)

width/height define unscaled dimensions; scaleX/scaleY scale to visual target size

The renderer auto-fits the actual image to cover the target area (width*scaleX by height*scaleY)

Circular image crop (optional):

{
  "type": "image",
  "name": "{{secondaryImage}}",
  "width": 300, "height": 300,
  "scaleX": 1, "scaleY": 1,
  "src": "data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAQAAAC1HAwCAAAAC0lEQVR42mNkYAAAAAYAAjCB0C8AAAAASUVORK5CYII=",
  "crossOrigin": "anonymous", "filters": [],
  "clipPath": {
    "type": "circle",
    "version": "5.3.0",
    "radius": 140,
    "left": 0, "top": 0,
    "originX": "center", "originY": "center",
    "fill": "rgb(0,0,0)",
    "inverted": false, "absolutePositioned": false
  }
}

#### Textbox Objects

Use large fontSize directly -- never use scaleX/scaleY to size text.

{
  "type": "textbox",
  "version": "5.3.0",
  "originX": "left", "originY": "top",
  "left": 70, "top": 1050,
  "width": 940, "height": 350,
  "fill": "rgba(255, 255, 255, 1)",
  "stroke": null, "strokeWidth": 0,
  "fontSize": 82, "fontFamily": "Arial", "fontWeight": 900,
  "text": "Headline Goes Here in Bold White Text",
  "textAlign": "left", "lineHeight": 1.08, "charSpacing": 0,
  "scaleX": 1, "scaleY": 1,
  "angle": 0, "opacity": 1, "visible": true,
  "name": "{{title}}",
  "selectable": true, "hasControls": true, "editable": true
}

Key rules:

fontSize MUST be the actual desired size. Do NOT use scaleX/scaleY to enlarge text -- this breaks auto-fit.

scaleX and scaleY MUST be 1 on text objects.

height defines the maximum bounding box -- set it generously. The renderer shrinks text to fit.

width defines the text wrapping width.

Recommended font sizes by role:

| Role | fontSize | fontWeight | Notes | |------|----------|------------|-------| | Title / Headline | 62-82 | 700-900 | Bold, high-impact | | Description / Body | 32-42 | 400 | Regular weight | | Label / Tag | 24-28 | 700 | Often with charSpacing: 80-150 | | Website / URL | 22-26 | 400-600 | Subtle, lower opacity | | CTA | 24-28 | 600-700 | Right-aligned or centered |

#### Gradient Overlays

Gradient overlays create dark zones over images where text remains readable. Use a 400x400 base rect, scale it up with scaleX/scaleY, and flip vertically with flipY: true so the solid color sits at the bottom.

{
  "type": "rect",
  "version": "5.3.0",
  "originX": "left", "originY": "top",
  "left": -20, "top": 540,
  "width": 400, "height": 400,
  "fill": {
    "type": "linear",
    "coords": {"x1": 200, "y1": 0, "x2": 200, "y2": 400},
    "colorStops": [
      {"offset": 0, "color": "rgba(0, 0, 0, 1)"},
      {"offset": 0.6, "color": "rgba(0, 0, 0, 0.75)"},
      {"offset": 0.9, "color": "rgba(0, 0, 0, 0)"}
    ],
    "offsetX": 0, "offsetY": 0, "gradientUnits": "pixels"
  },
  "stroke": null, "strokeWidth": 0,
  "scaleX": 2.85, "scaleY": 3.5,
  "flipY": true,
  "opacity": 1, "visible": true,
  "name": "gradient_overlay_1", "selectable": true
}

Best practices:

Stack TWO gradient rects with slightly different positions and color stop distributions for visual depth

flipY: true makes offset 0 (solid) appear at the bottom, offset 1 (transparent) at the top

Gradient covers the bottom 60-70% of the canvas; text sits in the fully opaque zone

Use left: -20 to ensure full-width coverage after scaling

Template Metadata

Describes which objects are dynamic and how they appear in the editor:

{
  "fields": [
    {
      "key": "title",
      "objectName": "{{title}}",
      "label": "Title",
      "maxLength": 80,
      "required": true,
      "defaultValue": "Headline Goes Here in Bold White Text"
    }
  ],
  "imageSlots": [
    {
      "key": "primaryImage",
      "objectName": "{{primaryImage}}",
      "required": true
    }
  ]
}

Rules:

Every {{key}} object in canvas JSON MUST have a matching entry in fields (text) or imageSlots (images)

objectName MUST exactly match the object's name including {{ and }}

Maximum 20 fields, maximum 10 image slots

defaultValue should match the text property of the corresponding textbox

Standard variable names:

| Key | Type | Purpose | Typical maxLength | |-----|------|---------|-------------------| | title | field | Main headline | 60-100 | | description | field | Supporting text | 150-250 | | label | field | Category tag (e.g., "BREAKING NEWS") | 20-30 | | website | field | URL or domain | 30 | | cta | field | Call to action (e.g., "Read Below") | 20-30 | | primaryImage | imageSlot | Main hero image | - | | secondaryImage | imageSlot | Supporting image (headshot, logo) | - |

API Reference

#### POST /templates -- Create Template

{
  "name": "News Card Dark",
  "description": "Dark cinematic news card with full-bleed hero image...",
  "category": "linkedin",
  "canvasJson": "{...stringified canvas JSON...}",
  "templateMetadata": {
    "fields": [...],
    "imageSlots": [...]
  }
}

Validation:

name: 1-200 characters, required

description: max 1000 characters, optional

category: "linkedin" | "instagram" | "article", required

canvasJson: stringified JSON, max 500KB, must contain a clip object

templateMetadata.fields: max 20 items

templateMetadata.imageSlots: max 10 items

Response (201):

{
  "success": true,
  "data": {
    "template": {
      "id": "uuid",
      "name": "News Card Dark",
      "description": "...",
      "category": "linkedin",
      "template_metadata": {...},
      "thumbnail_url": null,
      "created_at": "2026-04-10T..."
    }
  }
}

#### GET /templates -- List Templates

Query: ?category=linkedin&limit=10&offset=0

Returns lightweight metadata (no canvas_json). Use GET by ID for the full template.

{
  "success": true,
  "data": {
    "templates": [...],
    "pagination": { "total": 12, "limit": 10, "offset": 0, "hasMore": true }
  }
}

#### GET /templates/{id} -- Get Template

Returns full template including canvas_json.

#### PATCH /templates/{id} -- Update Template

Partial updates. Send only the fields to change (name, category, canvasJson, templateMetadata, isActive).

#### DELETE /templates/{id} -- Delete Template

Returns { "success": true, "data": { "deleted": true } }.

#### POST /templates/{id}/render -- Render Template (20 credits)

{
  "fields": {
    "title": "Breaking: AI Reaches New Milestone",
    "description": "Researchers announce breakthrough in reasoning capabilities.",
    "label": "TECH NEWS",
    "website": "technews.com"
  },
  "images": {
    "primaryImage": "https://example.com/hero.jpg"
  },
  "outputFormat": "jpeg",
  "quality": 0.9
}

fields: key-value pairs matching template field keys (all strings)

images: key-value pairs matching template image slot keys (HTTPS URLs only)

outputFormat: "jpeg" (default) or "png"

quality: 0.1-1.0, default 0.9

Response:

{
  "success": true,
  "data": {
    "imageUrl": "https://generatebot.b-cdn.net/posts/...",
    "width": 1080,
    "height": 1920,
    "creditsConsumed": 20,
    "pipelineRunId": "uuid"
  }
}

Text Auto-Fit

The renderer automatically shrinks text to fit within its bounding box: 1. Set height generously on textbox objects -- this is the max vertical space 2. Use the actual desired fontSize -- the renderer shrinks if text is longer than placeholder 3. Do not rely on scaleX/scaleY for text sizing -- keep both at 1 4. Test with long text to verify auto-fit behavior

Image Smart-Fit

The renderer scales images to cover their target area without distortion:

Target area = width * scaleX by height * scaleY

Image is scaled uniformly to fill (cover) the target area, centered

Just provide the image URL at render time -- the system handles all sizing

Design Patterns

Pattern 1: Hero Image with Gradient Overlay (Dark Theme) Best for news, dramatic stories. Full-bleed hero image, two stacked gradient rects, bold white title, subtle description, label at top, website/CTA at bottom.

Pattern 2: Square Card with Accent Line Best for feeds. 1080x1080, aggressive gradient, thin vertical accent line, large caps title, no description.

Pattern 3: Editorial Clean (Light Theme) Best for thought leadership. Light background, colored label badge, serif title, hero image in bottom half.

Pattern 4: Split Dual Image Best for comparisons. Two images separated by a colored title band.

Critical Rules

1. Image src must NEVER be empty string. Always use: data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAQAAAC1HAwCAAAAC0lEQVR42mNkYAAAAAYAAjCB0C8AAAAASUVORK5CYII= 2. Text scaleX and scaleY must be 1. Use fontSize for sizing. 3. The clip rect must be the first object with name: "clip". 4. Canvas-level clipPath must match the clip rect's dimensions and fill. 5. Image objects must include crossOrigin: "anonymous" and filters: []. 6. Gradient overlays use 400x400 base size scaled via scaleX/scaleY. Never use full canvas dimensions directly. 7. Stack TWO gradient rects for depth -- single gradients look flat. 8. Set generous height on textboxes -- this is the auto-fit bounding box. 9. Category must be "linkedin", "instagram", or "article". 10. canvasJson max size is 500KB. 11. Every {{key}} object must have a matching metadata entry.

Available Fonts

Arial, Arial Black, Verdana, Helvetica, Tahoma, Trebuchet MS, Times New Roman, Georgia, Garamond, Courier New, Brush Script MT, Palatino, Bookman, Comic Sans MS, Impact, Lucida Sans Unicode, Geneva, Lucida Console

Example: Create and Render a Template

Step 1: Create

curl -X POST https://generatebot.com/api/v1/templates \
  -H "Authorization: Bearer GENERATEBOT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "News Card Dark",
    "description": "Dark cinematic news card",
    "category": "linkedin",
    "canvasJson": "{...stringified canvas JSON...}",
    "templateMetadata": {
      "fields": [
        {"key": "title", "objectName": "{{title}}", "label": "Title", "maxLength": 80, "required": true, "defaultValue": "Your Headline"}
      ],
      "imageSlots": [
        {"key": "primaryImage", "objectName": "{{primaryImage}}", "required": true}
      ]
    }
  }'

Step 2: Render (20 credits)

curl -X POST https://generatebot.com/api/v1/templates/TEMPLATE_ID/render \
  -H "Authorization: Bearer GENERATEBOT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "fields": { "title": "AI Breakthrough Announced Today" },
    "images": { "primaryImage": "https://example.com/hero.jpg" },
    "outputFormat": "jpeg",
    "quality": 0.9
  }'

Error Handling

All errors: { "error": { "code": "ERROR_CODE", "message": "..." }, "requestId": "..." }

Other GenerateBot Skills

Core Skill (generatebot-core): Search, pipeline, RSS, content CRUD, and credits endpoints.

Video Skill (generatebot-video): Detailed video creation tutorial with imageMode reference.

Publish Skill (generatebot-publish): Social posting, CMS publishing, enrichment, brand voice, AI images.

Workflows Skill (generatebot-workflows): End-to-end workflow patterns and usage mapping.

GenerateBot Workflows - Reference Guide

Base URL: https://generatebot.com/api/v1

Authentication: Bearer token in the Authorization header.

Authorization: Bearer GENERATEBOT_API_KEY

All request and response bodies use JSON. Set Content-Type: application/json.

Credit Cost Summary

| Action | Credits | Skill | |--------|---------|-------| | Search (news-aggregator) | 10 | Core | | RSS feeds (all operations) | 0 | Core | | Content pipeline (content-analyzer) | 100 | Core | | Topic pipeline (full-pipeline) | 100 | Core | | Script generator pipeline | 15 | Core | | Video reel | 200 | Video | | Template render | 20 | Publish | | Enrich (links or images) | 20 each | Publish | | Brand voice rewrite | 5 | Publish | | AI image generation | 10 | Publish | | Reddit replies | 10 | Publish | | Business profile | 5 | Publish | | Save/list content | 0 | Core | | Credit transactions | 0 | Core | | Template CRUD | 0 | Publish | | Social media post | 0 | Publish | | Social post history | 0 | Publish | | CMS publish | 0 | Publish | | CMS publish history | 0 | Publish | | Check credits | 0 | Core |

Usage Pattern Mapping

| User Request | Actions | Credits | |-------------|---------|---------| | "search for [topic]" | POST /agents/news-aggregator | 10 | | "get RSS news" | POST /rss | 0 | | "write an article about [topic]" | POST /pipelines (full-pipeline) | 100 | | "create content from [url]" | POST /pipelines (content-analyzer) | 100 | | "make a video about [topic]" | Pipeline (100) + Video (200) | 300 | | "enrich with links/images" | POST /enrich/* | 20 each | | "generate images" | POST /agents/ai-image-generator | 10 | | "rewrite in brand voice" | POST /agents/brand-voice-rewriter | 5 | | "create a post template" | POST /templates | 0 | | "render a template" | POST /templates/{id}/render | 20 | | "post to [platform]" | POST /social/post (needs contentId) | 0 | | "publish to [cms]" | POST /cms/publish (needs contentId) | 0 | | "show post history" | GET /social/history or GET /cms/history | 0 | | "check credits" / "transactions" | GET /credits or GET /credits/transactions | 0 |

Complete Workflow: Search -> Article -> Video -> Publish

Step 1: Check credits -- GET /api/v1/credits (0 credits)

Confirm you have at least 310 credits for the full flow

Step 2: Search for a topic -- POST /api/v1/agents/news-aggregator (10 credits)

Send: { "topic": "your topic here" }

Extract the best article URL from the response array

Step 3: Generate content -- POST /api/v1/pipelines (100 credits)

Send: { "pipelineType": "content-analyzer", "url": "" }

Poll GET /api/v1/pipelines/{runId} every 10s until "completed"

Step 4: Extract content from completed pipeline

Article: results.data.agents.aiContentGenerator.generatedContent.articles[0]

Script: results.data.agents.scriptGenerator.scripts[0]

Images: results.data.agents.imageFinder.foundImages

Carousel: results.data.agents.carouselGenerator.carousel.slides

Step 5: Save the article -- POST /api/v1/content (0 credits)

{
  "contentType": "article",
  "title": "",
  "contentData": {
    "introduction": "",
    "sections": "",
    "metaDescription": ""
  }
}

Save the returned contentId at response.content.id.

Step 6: Publish article to CMS -- POST /api/v1/cms/publish (0 credits)

GET /api/v1/cms/accounts to get the cmsAccountId

Convert article to HTML:

introduction
heading
content

Send with contentId and cmsAccountId

Step 7: Create video reel -- POST /api/v1/videos (200 credits)

Build slides from carousel + images (see Video skill for full tutorial)

Include sourcePipelineRunId from step 3

Poll GET /api/v1/videos/{jobId} every 5s until "completed"

Step 8: Post video to social media -- POST /api/v1/social/post (0 credits)

GET /api/v1/social/accounts to get connected account IDs

Use the contentId from the video response with contentType: "video"

Total: ~310 credits (10 search + 100 pipeline + 200 video). CMS and social posting are free.

Tip: Use RSS feeds (POST /rss, 0 credits) instead of the news aggregator (10 credits) to source article URLs for free. Save your feed URLs once with PUT /rss, then fetch with { "useSavedFeeds": true }.

RSS-Based Workflow (Free Content Sourcing)

1. Save RSS feeds: PUT /api/v1/rss with { "feedUrls": ["https://techcrunch.com/feed/"] } 2. Fetch latest: POST /api/v1/rss with { "useSavedFeeds": true, "keywords": ["AI"], "maxItemsPerFeed": 5 } 3. Pick best article URL from the results 4. Run pipeline: POST /api/v1/pipelines on the chosen URL (100 credits) 5. Continue with save/publish/video as needed

Behavioral Guidelines

For operations costing 50+ credits (pipelines at 100, videos at 200), confirm the action and credit cost before proceeding. For cheap or free operations, proceed directly.

Show real-time progress while polling async operations. Report the current status on each poll cycle.

When a pipeline completes, summarize: article title, number of social posts, images found, carousel slides, and script availability.

When creating a video from pipeline results, always include sourcePipelineRunId.

Pipeline results are NOT auto-saved. Remind the user to save content (POST /content) before publishing.

If the user asks to "do everything" or "full workflow", follow the Complete Workflow above.

Polling Guide

Pipelines: Poll every 10 seconds.

Status: "pending" -> "running" -> "completed" | "failed"

Typical: 2-3 minutes. Max: 5 minutes.

Videos: Poll every 5 seconds.

Status: "queued" -> "generating_audio" -> "rendering" -> "uploading" -> "completed" | "failed"

"queued" can last 10-30 seconds (normal startup)

"generating_audio" means TTS in progress (progress, not a stall)

"rendering" is the longest phase

Typical: 60-180 seconds. Max: 5 minutes.

Images: Synchronous but slow (5-15 seconds per image). Wait for the response.

CRITICAL: Do NOT stop polling on intermediate statuses. Do NOT report failure until status is actually "failed" or you have polled for the maximum duration.

Error Handling

All errors: { "error": { "code": "ERROR_CODE", "message": "..." }, "requestId": "..." }

Other GenerateBot Skills

Core Skill (generatebot-core): All search, pipeline, RSS, content CRUD, and credits endpoints.

Video Skill (generatebot-video): Detailed video creation tutorial with imageMode reference.

Publish Skill (generatebot-publish): Social posting, CMS publishing, enrichment, brand voice, AI images.

Templates Skill (generatebot-templates): Design and render canvas-based post image templates.

⚙️ Configuration

Before creating a video, you need completed pipeline results. Run a content pipeline first (see Core skill):

POST /api/v1/pipelines
{ "pipelineType": "content-analyzer", "url": "https://example.com/article" }

Poll until status is "completed", then extract these three things:

1. Carousel slides: results.data.agents.carouselGenerator.carousel.slides

   [{ "slideNumber": 1, "text": "Key takeaway from the article" }, ...]

2. Images: results.data.agents.imageFinder.foundImages

   [{ "imageUrl": "https://...", "altText": "...", "suggestedUse": "hero" }, ...]

3. Hook: results.data.agents.scriptGenerator.scripts[0].hook

   "Breaking news you need to know!"

4. Pipeline Run ID: The pipelineRunId from the pipeline POST response.

Step 1: Build Video Slides

Each carousel slide becomes one video slide. Map them like this:

{
  "text": "",
  "imageUrl": "",
  "highlight": ["key", "words"]
}

Pairing rules:

Use the suggestedUse: "hero" image for slide 1

Use remaining images for subsequent slides

Each slide MUST have a unique imageUrl -- never reuse the same image across slides

text is max 200 characters

The text field is both the on-screen caption AND the voiceover narration. Write it as spoken language:

Numbers in words: "25 million dollars" not "$25M"

Use "A.I." not "AI" (TTS pronounces it better)

Add ellipsis for dramatic pauses: "And then... everything changed."

Step 2: Image Display Modes

The imageMode field controls how images appear in the 9:16 portrait frame:

Recommendation: Omit imageMode in most cases. Auto-detection handles landscape vs portrait correctly. Only set it explicitly to override.

Step 3: Extra Images and Word-Synced Timing

Add related images as overlays that appear synced to specific words in the narration:

{
  "text": "A stunning property in Brisbane with a resort style pool",
  "imageUrl": "https://cdn.example.com/mansion-exterior.jpg",
  "extraImages": [
    "https://cdn.example.com/mansion-aerial.jpg",
    "https://cdn.example.com/mansion-pool.jpg"
  ],
  "extraImageTimings": [
    { "showAtWordIndex": 2 },
    { "showAtWordIndex": 7 }
  ]
}

How word indexing works: Count words in text starting from 0, split on whitespace. Punctuation stays attached.

For the text above: A(0) stunning(1) property(2) in(3) Brisbane(4) with(5) a(6) resort(7) style(8) pool(9)

The aerial shot appears at word 2 ("property")

The pool shot appears at word 7 ("resort")

Each overlay stays visible until the next one starts or the slide ends

When to use extraImages:

Landscape images that lose detail when cropped to 9:16

Multiple angles of the same subject (exterior + pool + interior)

You CAN reuse the main imageUrl in extraImages for blurred-bg + clear-overlay effect

Limits: Max 5 extra images per slide. If you omit extraImageTimings, overlays distribute evenly (less precise).

Step 4: Highlights

The highlight field is an array of words from text to emphasize visually (rendered in accent color with glow effect):

"highlight": ["25 million", "mansion"]

Pick 1-3 key words or phrases that appear in the text

Max 10 items

If omitted, the API auto-generates 2 highlights (prefers numbers and capitalized words)

Step 5: Per-Slide Style Overrides (Optional)

Each slide can override global styles:

Advanced text styling (textStyle object, all fields optional):

{
  "textStyle": {
    "fontFamily": "Inter",
    "letterSpacing": 1.2,
    "lineHeight": 1.4,
    "textTransform": "uppercase",
    "shadow": { "enabled": true, "offsetX": 2, "offsetY": 2, "blur": 4, "color": "#000000" },
    "outline": { "enabled": true, "width": 2, "color": "#000000" },
    "background": { "enabled": true, "color": "#000000", "opacity": 0.7, "paddingX": 12, "paddingY": 8, "borderRadius": 8 },
    "glow": { "enabled": true, "blur": 10, "color": "#75F30F", "intensity": 0.8 }
  }
}

textTransform: "none", "uppercase", "lowercase", "capitalize"

shadow: Drop shadow behind text

outline: Stroke around text characters

background: Colored box behind text (like a caption box)

glow: Colored glow effect around text

Step 6: Set the Hook

"hook": ""

Max 200 characters

This is the opening text shown before slides begin

No separate script field needed -- the API auto-builds TTS narration from all slide texts

Step 7: Configure Global Style (Optional)

"style": {
  "accentColor": "#FF5500",
  "captionStyle": "outlined",
  "captionPosition": "center",
  "fontSize": 48,
  "colorGrade": "cinematic",
  "filmGrain": { "enabled": true, "opacity": 0.025, "fps": 8 }
}

Step 8: Configure Voice (Optional)

Step 9: Assemble and Submit

POST /api/v1/videos (200 credits)

{
  "hook": "$25M mansion hits the market!",
  "slides": [
    {
      "text": "A 25 million dollar mansion in Brisbane is up for grabs.",
      "imageUrl": "https://cdn.example.com/mansion-exterior.jpg",
      "highlight": ["25 million", "mansion"],
      "extraImages": ["https://cdn.example.com/mansion-aerial.jpg"],
      "extraImageTimings": [{ "showAtWordIndex": 4 }]
    },
    {
      "text": "Spanning 1400 square meters with five bedrooms and eight bathrooms.",
      "imageUrl": "https://cdn.example.com/mansion-pool.jpg",
      "highlight": ["1400", "bathrooms"]
    }
  ],
  "sourcePipelineRunId": "",
  "style": {
    "accentColor": "#75F30F",
    "captionStyle": "outlined",
    "captionPosition": "center"
  }
}

Field constraints:

hook: 1-200 chars, required

slides: 1-15 slides, required (4-7 recommended)

slides[].text: 1-200 chars, required

slides[].imageUrl: HTTPS URL, required

slides[].highlight: max 10 items, optional

slides[].extraImages: max 5 URLs, optional

watermark: small text in corner, max 100 chars, optional

watermarkLogoUrl: HTTPS URL to a logo image for the watermark, optional

watermarkPosition: position string (max 50 chars), optional

cta: { "text": "Follow for more!", "url": "example.com" }, optional

sourcePipelineRunId: UUID, strongly recommended -- links video to source content

Response:

{
  "jobId": "uuid",
  "contentId": "uuid",
  "status": "queued",
  "creditsConsumed": 200,
  "statusUrl": "/api/v1/videos/{jobId}"
}

Step 10: Poll for Completion

GET /api/v1/videos/{jobId} every 5 seconds.

Status progression: queued -> generating_audio -> rendering -> uploading -> completed

queued: 10-30s while render worker starts. Normal.

generating_audio: TTS is being generated. Progress, not a stall.

rendering: Video frames being composed. Longest phase.

completed: videoUrl contains the download URL.

failed: error field contains the error message.

Keep polling up to 5 minutes. Do NOT stop on intermediate statuses. Do NOT report failure until status is "failed" or you have polled the maximum duration.

Step 11: Post the Video

The video response includes a contentId. Use it directly with POST /api/v1/social/post (see Publish skill):

{
  "contentId": "",
  "platforms": [{
    "accountId": "",
    "platform": "instagram",
    "contentType": "video",
    "caption": "Check out our latest video!"
  }]
}

Render Time Estimates

| Slides | Approximate Time | |--------|------------------| | 3-5 | 60-90 seconds | | 6-10 | 90-150 seconds | | 11-15 | 150-180 seconds |

Audio generation adds 10-20 seconds. Upload adds 5-10 seconds.

Error Handling

Other GenerateBot Skills

Core Skill (generatebot-core): Search for content, run pipelines, manage content library. Required before creating videos.

Publish Skill (generatebot-publish): Post completed videos to social media, publish articles to CMS.

Templates Skill (generatebot-templates): Design and render canvas-based post image templates.

Workflow Skill (generatebot-workflows): End-to-end workflow examples and patterns.