Media Collections

Overview

Two primary collections manage creator media content: scraped_media for content with engagement metrics, and scheduled_posts for the publishing queue.

scraped_media

Stores creator content scraped from platform accounts, including engagement metrics, taxonomy classifications, and media metadata.

Purpose

Archive all scraped content from connected platforms
Track engagement metrics (likes, comments, views, revenue)
Store taxonomy classifications for content discovery
Link media to creator profiles and platform sources
Enable AI-powered content analysis and recommendations

Key Fields

Field	Type	Description
`id`	UUID	Primary key
`creator_profile_id`	Foreign Key	Links to `creator_profiles`
`platform`	String	Source platform (“onlyfans”, “fansly”, etc.)
`platform_media_id`	String	Platform’s unique identifier for this content
`media_type`	String	`image`, `video`, `gallery`, `text`
`media_url`	String	Original media URL or local storage path
`thumbnail_url`	String	Thumbnail/preview image URL
`caption`	Text	Original caption/description
`posted_at`	DateTime	When content was posted on platform
`likes_count`	Integer	Number of likes
`comments_count`	Integer	Number of comments
`views_count`	Integer	View count
`revenue`	Decimal	Revenue generated (if available)
`taxonomy_tags`	JSON	Array of taxonomy dimension assignments
`scraped_at`	DateTime	When this was scraped
`created_at`	DateTime	Record creation timestamp

Taxonomy Integration

Media items are classified using the 6-concept taxonomy system:

// taxonomy_tags field example
{
  "body_type": ["athletic", "curvy"],
  "performance_style": ["sensual", "playful"],
  "setting": ["bedroom", "outdoor"],
  "attire": ["lingerie", "casual"],
  "audience_appeal": ["mainstream", "fetish"],
  "production_quality": ["professional"]
}

See Taxonomy Collections for full classification system details.

Example Queries

List Recent Media

await use_mcp_tool({
  server_name: "directus",
  tool_name: "read-items",
  arguments: {
    collection: "scraped_media",
    fields: ["id", "media_type", "caption", "likes_count", "posted_at"],
    sort: ["-posted_at"],
    limit: 20
  }
});

Find Top Performing Content

await use_mcp_tool({
  server_name: "directus",
  tool_name: "read-items",
  arguments: {
    collection: "scraped_media",
    fields: ["id", "caption", "likes_count", "revenue", "media_type"],
    filter: {
      media_type: { _eq: "video" },
      posted_at: { _gte: "$NOW(-30 days)" }
    },
    sort: ["-likes_count"],
    limit: 10
  }
});

Search by Taxonomy Tags

await use_mcp_tool({
  server_name: "directus",
  tool_name: "search-items",
  arguments: {
    collection: "scraped_media",
    query: "athletic",
    fields: ["id", "caption", "taxonomy_tags", "media_type"]
  }
});

scheduled_posts

Post publishing queue polled every 60 seconds by the media worker.

Purpose

Queue posts for cross-platform publishing
Schedule posts for specific dates/times
Track publishing status and errors
Support multi-platform simultaneous posting

Key Fields

Field	Type	Description
`id`	UUID	Primary key
`creator_profile_id`	Foreign Key	Links to `creator_profiles`
`media_id`	Foreign Key	Optional link to `scraped_media`
`platforms`	JSON	Array of platform targets `["onlyfans", "fansly"]`
`caption`	Text	Post caption/description
`media_urls`	JSON	Array of media file URLs to publish
`scheduled_for`	DateTime	When to publish (null = immediate)
`status`	String	`pending`, `processing`, `published`, `failed`
`published_at`	DateTime	Actual publish timestamp
`error_message`	Text	Error details if status=failed
`retry_count`	Integer	Number of publish attempts
`created_at`	DateTime	Queue entry creation timestamp

Polling Mechanism

The media worker (media-worker/index.js) polls this collection every 60 seconds:

// Worker logic (simplified)
setInterval(async () => {
  const posts = await directus.items('scheduled_posts').readByQuery({
    filter: {
      status: 'pending',
      scheduled_for: { _lte: new Date() }
    }
  });
  
  for (const post of posts) {
    await publishQueue.add('publish_post', { postId: post.id });
  }
}, 60000);

Example Queries

Create Scheduled Post

await use_mcp_tool({
  server_name: "directus",
  tool_name: "create-item",
  arguments: {
    collection: "scheduled_posts",
    data: {
      creator_profile_id: "profile-uuid",
      platforms: ["onlyfans", "fansly"],
      caption: "New content dropping tonight!",
      media_urls: ["/uploads/video123.mp4"],
      scheduled_for: "2026-03-05T20:00:00Z",
      status: "pending"
    }
  }
});

List Pending Posts

await use_mcp_tool({
  server_name: "directus",
  tool_name: "read-items",
  arguments: {
    collection: "scheduled_posts",
    fields: ["id", "caption", "platforms", "scheduled_for", "status"],
    filter: {
      status: { _in: ["pending", "processing"] }
    },
    sort: ["scheduled_for"]
  }
});

Update Post Status

await use_mcp_tool({
  server_name: "directus",
  tool_name: "update-item",
  arguments: {
    collection: "scheduled_posts",
    id: "post-uuid",
    data: {
      status: "published",
      published_at: new Date().toISOString()
    }
  }
});

creator_profiles - Platform accounts that create/publish content
media_jobs - Background processing jobs for media operations
taxonomy_mapping - Tag classifications applied to media

Workflow Integration

Content Scraping → Storage Flow

Media worker runs scrape_profile job
Stagehand extracts content from platform
Records created in scraped_media with engagement metrics
AI taxonomy classification applied (optional)
Content available in Media Library dashboard

Post Publishing Flow

User creates post via /app/calendar or AI chat
Record created in scheduled_posts with status=pending
Worker polls every 60s for posts where scheduled_for <= NOW
Creates media_jobs entry with type publish_post
Stagehand browser automation publishes to target platforms
Status updated to published or failed with error details

Best Practices

Use scheduled_for for timing - Set to current time for immediate publish, future time for scheduling
Monitor retry_count - Posts failing 3+ times may need manual intervention
Clean up old media - Archive or delete scraped_media after 90+ days to manage storage
Batch taxonomy classification - Use taxonomy-tag action flow to classify multiple items
Handle platform differences - Some platforms may require platform-specific caption formats

Endpoints

MCP Tools

Collections

Overview

scraped_media

Purpose

Key Fields

Taxonomy Integration

Example Queries

List Recent Media

Find Top Performing Content

Search by Taxonomy Tags

scheduled_posts

Purpose

Key Fields

Polling Mechanism

Example Queries

Create Scheduled Post

List Pending Posts

Update Post Status

Workflow Integration

Content Scraping → Storage Flow

Post Publishing Flow

Best Practices

See Also

Build docs developers (and LLMs) love

Endpoints

MCP Tools

Collections

​Overview

​scraped_media

​Purpose

​Key Fields

​Taxonomy Integration

​Example Queries

​List Recent Media

​Find Top Performing Content

​Search by Taxonomy Tags

​scheduled_posts

​Purpose

​Key Fields

​Polling Mechanism

​Example Queries

​Create Scheduled Post

​List Pending Posts

​Update Post Status

​Related Collections

​Workflow Integration

​Content Scraping → Storage Flow

​Post Publishing Flow

​Best Practices

​See Also

Build docs developers (and LLMs) love

Overview

scraped_media

Purpose

Key Fields

Taxonomy Integration

Example Queries

List Recent Media

Find Top Performing Content

Search by Taxonomy Tags

scheduled_posts

Purpose

Key Fields

Polling Mechanism

Example Queries

Create Scheduled Post

List Pending Posts

Update Post Status

Related Collections

Workflow Integration

Content Scraping → Storage Flow

Post Publishing Flow

Best Practices

See Also