PowerPost Skill

PowerPost writes social media captions, generates images and videos, and publishes to Instagram, TikTok, X/Twitter, YouTube, Facebook, and LinkedIn through a single API.

Setup

The user needs two credentials:

• POWERPOST_API_KEY — their API key (starts with pp_live_sk_). They can create one at https://powerpost.ai/settings/api
• POWERPOST_WORKSPACE_ID — their workspace ID. Found at https://powerpost.ai/settings/workspaces

There are two ways to configure these in OpenClaw. The API key can go in the Skills UI (the "API key" field) or via config. The workspace ID has to be set via config since the UI only has one field per skill:

openclaw config set skills.entries.powerpost.apiKey "pp_live_sk_YOUR_KEY"
openclaw config set skills.entries.powerpost.env.POWERPOST_WORKSPACE_ID "YOUR_WORKSPACE_ID"

API keys come in two types:

• read_write — full access, including publishing.
• read_draft — can generate content and create drafts, but publishing is blocked (returns 403). Good when a human should review before anything goes live.

New accounts start with 50 free credits. Pricing details at https://powerpost.ai/pricing

First run

Before running any PowerPost command, make sure both credentials are set. If either is missing, walk the user through setup:

1. If POWERPOST_API_KEY is not set:

   • Ask if they have a PowerPost account. If not, they can sign up at https://powerpost.ai/login
   • Point them to https://powerpost.ai/settings/api to create an API key.
   • They can paste it into the "API key" field on the Skills page in the OpenClaw UI, or run: openclaw config set skills.entries.powerpost.apiKey "pp_live_sk_..."

2. If POWERPOST_WORKSPACE_ID is not set:

   • Point them to https://powerpost.ai/settings/workspaces to find it.
   • They need to run: openclaw config set skills.entries.powerpost.env.POWERPOST_WORKSPACE_ID "their-workspace-id" — there's no UI field for this one, it has to be set via config.

3. Once both are set, verify by calling GET /account/credits. If it returns a balance, you're good. If it returns 401, the API key is wrong.

Don't run any other PowerPost commands until both credentials are confirmed working.

Base URL

All requests go to:

https://powerpost.ai/api/v1

Authentication Headers

Every request needs x-api-key (or Authorization: Bearer <key>). All content endpoints also need X-Workspace-Id. The only exception is GET /account/credits which only needs x-api-key.

x-api-key: $POWERPOST_API_KEY
X-Workspace-Id: $POWERPOST_WORKSPACE_ID

Core Workflow

The standard PowerPost flow is:

1. Check credits to make sure the user has enough balance.
2. Generate content from text, images, or video (async — returns a generation ID).
3. Poll for results every 2-3 seconds until status is completed or failed.
4. Optionally generate images or videos from the captions or a text prompt (also async — poll for results).
5. Create a post combining the generated captions and images into a draft.
6. Show the draft to the user for review before publishing.
7. Publish the post to connected social platforms.

Always check credits before generating content. Always show the user what will be published and get confirmation before calling the publish endpoint.

Endpoints

1. Check Credits

Use this to check how many credits the user has before starting any generation.

curl https://powerpost.ai/api/v1/account/credits \
  -H "x-api-key: $POWERPOST_API_KEY"

Note: This endpoint does NOT need the X-Workspace-Id header.

Response:

{
  "balance": 47
}

Tell the user their current balance. If balance is low, warn them before they start a generation.

Credit Costs

Credit costs vary by action:

• Caption generation: depends on research mode, input type, number of source URLs, and video length.
• Image generation: depends on model and quantity.
• Video generation: depends on model, duration, and whether audio is enabled.
• Publishing: depends on platform. Premium platforms like X cost more.

The exact cost is returned in every API response via credits_used and remaining_credits. Always check credits before starting a generation so the user knows the cost upfront.

Credits are only charged for successful operations. Failed generations, failed publish attempts, and failed URL scrapes are refunded.

2. Upload Media

Use this when the user wants to generate content from images or video, or when they want to attach media to a post.

curl -X POST https://powerpost.ai/api/v1/media/upload \
  -H "x-api-key: $POWERPOST_API_KEY" \
  -H "X-Workspace-Id: $POWERPOST_WORKSPACE_ID" \
  -F "file=@/path/to/file" \
  -F "type=image"

The type field must be image or video.

Supported formats:

• Images: JPEG, PNG, WebP (max 10 MB)
• Videos: MP4, MOV, WebM (max 500 MB)

Response:

{
  "media_id": "a1b2c3d4-e5f6-7890-abcd-ef1234567890",
  "type": "image",
  "file_name": "photo.jpg",
  "file_size": 245000,
  "mime_type": "image/jpeg",
  "created_at": "2026-01-10T18:30:00Z"
}

Save the media_id from the response. You will need it for content generation (media_ids field) or for creating posts (media_ids in post items).

Tell the user the upload was successful and show the file name and size.

3. Generate Content

Use this when the user wants to create social media captions.

There are three input modes, determined by which fields you provide:

• Text only: Send prompt (required, 3-2000 chars).
• Image input: Send media_ids with image IDs (+ optional prompt for context).
• Video input: Send media_ids with a video ID (+ optional prompt for context).

You must provide either prompt or media_ids (or both).

curl -X POST https://powerpost.ai/api/v1/content/generate \
  -H "x-api-key: $POWERPOST_API_KEY" \
  -H "X-Workspace-Id: $POWERPOST_WORKSPACE_ID" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "We just shipped dark mode across all our apps",
    "post_types": ["instagram-reel", "tiktok-video", "x-post"],
    "research_mode": "regular"
  }'

Required fields:

• post_types (string array) — At least one post type. See the post types table below.
• research_mode (string) — "regular" or "deep". Use regular for quick tasks, deep for important posts.

Optional fields:

• writing_style_id (string) — A custom writing style ID created in the dashboard.
• cta_text (string, max 100 chars) — A custom call-to-action.
• source_urls (string array, max 10) — URLs to scrape for research context. Must be HTTPS. Private/internal URLs are filtered out for security. Each URL adds to the generation cost.

Post types:

Multiple post types from the same platform produce one shared caption for that platform. The post type declares publishing intent.

Response:

{
  "generation_id": "550e8400-e29b-41d4-a716-446655440000",
  "status": "processing",
  "credits_used": 10,
  "remaining_credits": 90,
  "status_url": "/api/v1/content/generations/550e8400-e29b-41d4-a716-446655440000"
}

Tell the user the generation has started, how many credits it cost, and that you are waiting for it to complete. Then immediately start polling.

4. Get Generation Status (Polling)

After starting a generation, poll this endpoint every 2-3 seconds until status is completed or failed. Do not poll more than 60 times (about 2 minutes).

curl https://powerpost.ai/api/v1/content/generations/GENERATION_ID \
  -H "x-api-key: $POWERPOST_API_KEY" \
  -H "X-Workspace-Id: $POWERPOST_WORKSPACE_ID"

While processing:

{
  "generation_id": "550e8400-...",
  "status": "processing",
  "prompt": "We just shipped dark mode",
  "platforms": ["tiktok", "instagram"],
  "research_mode": "regular",
  "credits_used": 10,
  "created_at": "2026-01-10T18:30:00Z"
}

If status is still processing or pending, wait 3 seconds and poll again.

When completed:

{
  "generation_id": "550e8400-...",
  "status": "completed",
  "prompt": "We just shipped dark mode",
  "platforms": ["tiktok", "instagram"],
  "research_mode": "regular",
  "credits_used": 10,
  "created_at": "2026-01-10T18:30:00Z",
  "outputs": {
    "tiktok": "Dark mode activated! POV: your eyes at 2am finally getting some relief... #darkmode #tech",
    "instagram": "Dark mode is here!\n\nYour late-night scrolling just got easier on the eyes... #DarkMode #ProductUpdate"
  }
}

The outputs object is keyed by platform name (not post type). Each value is a string, EXCEPT YouTube which has a title and description:

"youtube": {
  "title": "We Just Shipped Dark Mode",
  "description": "Dark mode is finally here across all our apps... #darkmode"
}

When completed, display all generated captions to the user, clearly labeled by platform.

When failed:

The response includes an error object with code and message. Credits are refunded on failure. Tell the user the generation failed and show the error message.

5. List Generations

Use this when the user wants to see their recent generations or search for a past generation.

curl "https://powerpost.ai/api/v1/content/generations?status=completed&limit=10" \
  -H "x-api-key: $POWERPOST_API_KEY" \
  -H "X-Workspace-Id: $POWERPOST_WORKSPACE_ID"

Query parameters (all optional):

• status — Filter by: pending, processing, completed, failed
• limit — Results per page, 1-100 (default 20)
• cursor — Pagination cursor from next_cursor in a previous response

Response:

{
  "data": [
    {
      "generation_id": "550e8400-...",
      "status": "completed",
      "prompt": "We just shipped dark mode",
      "platforms": ["tiktok", "instagram"],
      "research_mode": "regular",
      "credits_used": 10,
      "created_at": "2026-01-10T18:30:00Z"
    }
  ],
  "next_cursor": "gen_cursor_abc123",
  "has_more": true
}

The list endpoint does NOT include outputs. To get the full content of a specific generation, use the get generation endpoint with its ID.

Show the user a summary table of their generations. If they want to see the content of a specific one, fetch it with the get endpoint.

6. Generate Images

Use this when the user wants to create AI-generated images. There are three input modes:

Text mode: Describe the image in prompt. Reference mode: Provide prompt + style_images (media IDs of uploaded reference images, max 4). Post mode: Provide generation_id to create images matching an existing caption generation (+optional prompt for extra direction and source_post_type to pick which platform caption to use).

curl -X POST https://powerpost.ai/api/v1/images/generate \
  -H "x-api-key: $POWERPOST_API_KEY" \
  -H "X-Workspace-Id: $POWERPOST_WORKSPACE_ID" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "Minimalist workspace with laptop and coffee, warm morning light",
    "size": "square",
    "quantity": 2,
    "model": "flux2-flex"
  }'

Fields:

• prompt (string, max 2000 chars) — Required for text and reference modes. Optional for post mode.
• size (string) — square (1:1), feed (4:5), portrait (9:16), landscape (16:9). Default: square.
• quantity (number, 1-4) — How many images to generate. Default: 1.
• model (string) — flux2-flex (default, best for multi-reference and fine control), ideogram-3 (best text rendering and logos), gpt-image-1.5 (most photorealistic, best detail), nano-banana-2 (fast generation, great text).
• enhance_prompt (boolean) — Let AI optimize the prompt. Default: false.
• style_images (string array) — Media IDs of reference images (max 4). For reference mode.
• generation_id (string) — Caption generation ID. For post mode.
• source_post_type (string) — Which platform's caption to use as source. For post mode only.

Model and size compatibility:

• flux2-flex: square, feed, portrait, landscape
• ideogram-3: square, feed, portrait, landscape
• gpt-image-1.5: square, portrait, landscape (NO feed)
• nano-banana-2: square, feed, portrait, landscape

If the user requests a model/size combo that is not supported, warn them and suggest an alternative.

Response:

{
  "image_generation_id": "7a8b9c0d-e1f2-3456-abcd-ef7890123456",
  "status": "processing",
  "credits_used": 10,
  "remaining_credits": 36,
  "status_url": "/api/v1/images/generations/7a8b9c0d-e1f2-3456-abcd-ef7890123456"
}

Tell the user that image generation has started and begin polling.

7. Get Image Generation Status (Polling)

Poll this endpoint every 2-3 seconds until status is completed or failed.

curl https://powerpost.ai/api/v1/images/generations/IMAGE_GENERATION_ID \
  -H "x-api-key: $POWERPOST_API_KEY" \
  -H "X-Workspace-Id: $POWERPOST_WORKSPACE_ID"

When completed:

{
  "image_generation_id": "7a8b9c0d-...",
  "status": "completed",
  "prompt": "Minimalist workspace with laptop and coffee",
  "size": "square",
  "quantity": 2,
  "created_at": "2026-01-10T18:30:00Z",
  "images": [
    {
      "media_id": "img-001-abcd-efgh",
      "url": "https://powerpost.ai/storage/images/img-001-abcd-efgh.webp",
      "thumbnail_url": "https://powerpost.ai/storage/images/img-001-abcd-efgh-thumb.jpg",
      "width": 1024,
      "height": 1024
    }
  ]
}

Each image has a media_id which you use to attach the image to a post. The url is a preview link valid for 7 days. Always use the media_id, not the URL, when creating posts.

Show the user the image URLs so they can preview. Save the media_id values for creating posts.

Partial failure: If some images fail while others succeed, you get a partial result with a partial credit refund. The response status will be failed but may still contain an images array with the successful ones.

8. Generate Videos

Use this when the user wants to create AI-generated videos. There are two input modes:

Text-to-video: Describe the video in prompt. Image-to-video: Provide prompt + source_image (media ID of an uploaded image).

curl -X POST https://powerpost.ai/api/v1/videos/generate \
  -H "x-api-key: $POWERPOST_API_KEY" \
  -H "X-Workspace-Id: $POWERPOST_WORKSPACE_ID" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "A golden retriever running through a field of wildflowers at sunset",
    "size": "landscape",
    "duration": 5,
    "model": "kling-3.0-pro",
    "has_audio": false
  }'

Fields:

• prompt (string, required) — Text description of the video.
• size (string) — landscape (16:9) or portrait (9:16). Default: landscape.
• duration (integer, required) — Duration in seconds. Must match one of the model's supported values (see table).
• model (string) — kling-3.0-pro (default), veo-3.1, or runway-gen-4.5.
• has_audio (boolean) — Generate with audio. Default: false. Only supported by some models.
• enhance_prompt (boolean) — Let AI optimize the prompt. Default: false.
• source_image (media ID, UUID) — Media ID of an uploaded image for image-to-video mode.

Model details:

Prompt limits by model: Veo 3.1 = 3000 chars, Kling 3.0 Pro = 2500 chars, Runway Gen-4.5 = 1000 chars.

Credit costs depend on the model, duration, and audio. The exact cost is returned in the response.

Response:

{
  "video_generation_id": "7a8b9c0d-e1f2-3456-abcd-ef7890123456",
  "status": "processing",
  "credits_used": 15,
  "remaining_credits": 85,
  "status_url": "/api/v1/videos/generations/7a8b9c0d-e1f2-3456-abcd-ef7890123456"
}

Tell the user that video generation has started and begin polling.

9. Get Video Generation Status (Polling)

Poll this endpoint every 10 seconds until status is completed or failed. Video generation takes longer than images — do not poll more than 120 times.

curl https://powerpost.ai/api/v1/videos/generations/VIDEO_GENERATION_ID \
  -H "x-api-key: $POWERPOST_API_KEY" \
  -H "X-Workspace-Id: $POWERPOST_WORKSPACE_ID"

When completed:

{
  "video_generation_id": "7a8b9c0d-...",
  "status": "completed",
  "prompt": "A golden retriever running...",
  "size": "landscape",
  "duration": 5,
  "has_audio": false,
  "model": "kling-3.0-pro",
  "created_at": "2026-03-11T10:00:00Z",
  "video": {
    "media_id": "vid-001-abcd",
    "url": "https://powerpost.ai/storage/videos/vid-001-abcd.mp4",
    "thumbnail_url": "https://powerpost.ai/storage/thumbs/vid-001-abcd.jpg",
    "width": 1920,
    "height": 1080,
    "duration": 5
  }
}

The video has a media_id which you use to attach it to a post. The url is a preview link valid for 7 days. Always use the media_id, not the URL, when creating posts.

Show the user the video URL so they can preview. Save the media_id for creating posts.

When failed: The response includes an error object with code (VIDEO_GENERATION_FAILED) and message. Credits are refunded on failure.

10. Create Post

Use this to assemble captions and media into a draft post ready for review and publishing.

curl -X POST https://powerpost.ai/api/v1/posts \
  -H "x-api-key: $POWERPOST_API_KEY" \
  -H "X-Workspace-Id: $POWERPOST_WORKSPACE_ID" \
  -H "Content-Type: application/json" \
  -d '{
    "generation_id": "550e8400-e29b-41d4-a716-446655440000",
    "items": [
      {
        "post_type": "instagram-reel",
        "media_ids": ["img-001-abcd-efgh"]
      },
      {
        "post_type": "tiktok-video",
        "media_ids": ["img-001-abcd-efgh"]
      }
    ]
  }'

Fields:

• generation_id (string, optional) — Links to a completed caption generation. If provided, content is auto-filled from the generation outputs for each platform.
• items (array, required) — One item per post type you want to publish.

Each item:

• post_type (string, required) — The target post type (e.g., instagram-reel, x-post).
• content (string) — The caption text. Required if no generation_id is provided. If generation_id is provided, content is auto-filled but can be overridden.
• title (string) — Required for YouTube post types (youtube-video, youtube-short).
• media_ids (string array, optional) — Media to attach (uploaded or generated image IDs).

Custom content example (no generation):

curl -X POST https://powerpost.ai/api/v1/posts \
  -H "x-api-key: $POWERPOST_API_KEY" \
  -H "X-Workspace-Id: $POWERPOST_WORKSPACE_ID" \
  -H "Content-Type: application/json" \
  -d '{
    "items": [
      {
        "post_type": "x-post",
        "content": "Just shipped the biggest update of the year."
      },
      {
        "post_type": "linkedin-post",
        "content": "Excited to announce our biggest product update of the year..."
      }
    ]
  }'

Response:

{
  "post_id": "post-550e8400-e29b-41d4-a716-446655440000",
  "status": "draft",
  "created_at": "2026-01-10T18:30:00Z",
  "items": [
    {
      "item_id": "item-001",
      "post_type": "instagram-reel",
      "platform": "instagram",
      "content": "Dark mode is here! Your late-night scrolling just got easier... #DarkMode",
      "media_ids": ["img-001-abcd-efgh"],
      "status": "draft"
    }
  ]
}

After creating a post, show the user a summary of every item (platform, content preview, attached media) and ask if they want to publish or make changes.

11. Get Post

Use this to check the status of a post, especially after publishing.

curl https://powerpost.ai/api/v1/posts/POST_ID \
  -H "x-api-key: $POWERPOST_API_KEY" \
  -H "X-Workspace-Id: $POWERPOST_WORKSPACE_ID"

Response includes the post with all items and their individual statuses. After publishing, each item will show posting, posted, or failed with an optional platform_post_id for successfully published items.

Post statuses: draft, publishing, published, failed. Item statuses: draft, posting, posted, failed.

12. Publish Post

Use this to publish a draft post to connected social platforms. Always confirm with the user before calling this endpoint.

curl -X POST https://powerpost.ai/api/v1/posts/POST_ID/publish \
  -H "x-api-key: $POWERPOST_API_KEY" \
  -H "X-Workspace-Id: $POWERPOST_WORKSPACE_ID"

Prerequisites:

• The user's social platforms must be connected at https://powerpost.ai/settings/connections
• The post must be in draft status.
• The user must have sufficient credits.
• The API key must be read_write type (not read_draft).

Response:

{
  "post_id": "post-550e8400-...",
  "status": "publishing",
  "credits_used": 1,
  "status_url": "/api/v1/posts/post-550e8400-..."
}

Publishing is asynchronous. After receiving the response, poll the Get Post endpoint every 3 seconds to check when each item has finished publishing.

Publishing costs vary by platform. The exact cost is returned in the response via credits_used.

Tell the user how many credits were used and that publishing is in progress. When all items show posted or failed, summarize the results.

Common User Commands

"Post about X" / "Create a post about X"

Full flow:

1. Check credits.
2. Ask which platforms if not specified. Default to all if the user says "everywhere" or "all platforms".
3. Generate content with the user's topic as the prompt. Use regular research mode unless they ask for deep. If the user provides URLs as references, include them in source_urls.
4. Poll until complete.
5. Show the generated captions.
6. Ask if they want to generate images too.
7. Create the post.
8. Show the draft and ask for confirmation.
9. Publish on confirmation.

"Write captions about X" / "Generate content about X"

Partial flow (no publishing):

1. Check credits.
2. Generate content.
3. Poll until complete.
4. Show the captions to the user.

"Create an image" / "Generate an image of X"

1. Check credits.
2. If the user provides a text description, use text mode. If they reference a previous generation, use post mode. If they have reference images, use reference mode.
3. Ask about size preference if not specified (square is a safe default).
4. Generate the image.
5. Poll until complete.
6. Show the image URLs to the user.

"Create a video" / "Generate a video of X"

1. Check credits.
2. If the user has an image they want to animate, use image-to-video mode (upload first, then pass source_image). Otherwise use text-to-video mode.
3. Ask about duration if not specified. Default to 5 seconds.
4. Ask about size if not specified (landscape is a safe default for video).
5. Generate the video.
6. Poll until complete (video takes longer than images — up to a few minutes).
7. Show the video URL to the user.

"Check my credits" / "How many credits do I have?"

1. Call the credits endpoint.
2. Tell the user their balance.

"Upload this image/video"

1. Upload the file using the media upload endpoint.
2. Tell the user the upload succeeded and show the media ID.
3. Ask what they want to do with it (generate captions from it, use it in a post, etc.).

"Show my recent posts" / "List my generations"

1. Call the list generations endpoint.
2. Show a summary of recent generations.
3. If the user wants details on one, fetch it with the get generation endpoint.

"Publish my post" / "Send it live"

1. If there is a draft post from this session, publish it.
2. If not, ask the user which post to publish.
3. Always confirm before publishing.

Polling Strategy

Content generation, image generation, and video generation are all asynchronous. After starting any of them:

1. Wait 2 seconds.
2. Call the status endpoint.
3. If status is processing or pending, wait 3 seconds and poll again.
4. If status is completed, show the results.
5. If status is failed, show the error and tell the user credits were refunded.
6. Do not poll more than 60 times for content/images, or 120 times for video (it takes longer). If it times out, tell the user to try again later.

Publishing is also asynchronous. After calling publish:

1. Wait 3 seconds.
2. Poll the Get Post endpoint.
3. Check each item's status. When all are either posted or failed, stop polling.

Error Handling

All errors return this format:

{
  "error": {
    "message": "Human-readable error message",
    "code": "ERROR_CODE"
  }
}

Handle these errors:

For rate limit errors (429), the response includes a retryAfter field in seconds. Wait that long before retrying.

Every response includes an X-Request-Id header. If the user needs support, give them this ID.

Things to know

• Always show the user what will be posted and where, and get explicit confirmation before publishing. This goes to real social accounts.
• YouTube posts need a title field in addition to the caption.
• If the user picks both instagram-reel and instagram-feed, they share one Instagram caption. The post type controls the publishing format, not the caption.
• Image and video URLs from generation expire after 7 days. Always use media_id when creating posts, not the URL.
• Credits are shared across all workspaces on an account.
• Workspaces are isolated from each other. Generations, posts, media, and connections in one workspace aren't visible in another.
• Users can set up webhooks at https://powerpost.ai/settings/api for real-time notifications instead of polling. Not needed for agent usage, but useful for production integrations.

External endpoints

All requests go to one domain: https://powerpost.ai/api/v1/*. Nothing is sent anywhere else.

Security and privacy

• Your API key and workspace ID are sent to powerpost.ai as HTTP headers with every request. They're not logged, displayed, or sent anywhere else.
• Prompts, uploaded media, and generated content are stored on PowerPost's servers, scoped to your workspace.
• This skill only uses curl over HTTPS. It doesn't write files, run scripts, or touch your system.

Trust

When you use this skill, your prompts, images, videos, and credentials go to powerpost.ai over HTTPS. PowerPost uses third-party AI models on their servers to process your content. Their privacy policy is at https://powerpost.ai/terms and support is at https://powerpost.ai/contact

Tips

• Always check the credit balance before starting a generation. Better to warn the user upfront than hit an insufficient credits error mid-flow.
• Default to regular research mode. Only use deep when the user asks for it or says the post matters.
• Default to square for image size if the user doesn't say. It works on most platforms.
• Default to flux2-flex for the image model. It supports all sizes and handles reference images well.
• If a generation fails, credits are refunded. Let the user know and suggest trying again.
• When the user says "all platforms," use one post type per platform: instagram-feed, tiktok-video, youtube-short, x-post, facebook-post, linkedin-post.
• If the user's image description is short or vague, suggest setting enhance_prompt: true to let the AI fill in the details.
• Default to kling-3.0-pro for video generation. It handles both sizes and supports audio.
• Default to landscape and 5 seconds for video unless the user says otherwise.
• Video generation is slower than images. Tell the user it may take a couple minutes.
• If the user provides URLs as references for a post, pass them as source_urls in the content generation request. Each URL adds to the cost.