AutoEditor MCP Server

Model Context Protocol server that gives AI agents (Claude Code, OpenAI, Cursor, etc.) access to AutoEditor's video intelligence.

Quick Start

Install

npm install -g autoeditor-mcp

With Claude Code

Add to your Claude Code config (~/.config/claude-code/mcp.json):

{
  "mcpServers": {
    "autoeditor": {
      "command": "npx",
      "args": ["autoeditor-mcp"],
      "env": {
        "ATLAS_API_KEY": "your-b2b-api-key",
        "ATLAS_API_BASE": "https://autoeditor.app"
      }
    }
  }
}

With OpenAI / Custom Agents

Run as a stdio process:

ATLAS_API_KEY=your-key autoeditor-mcp

Tools

score_video_retention

Score a video's viewer retention probability before it goes live. Returns overall score (0-100), per-segment breakdown, predicted retention curve, and top issues.

get_retention_advice

Get concrete, timestamped recommendations for improving a video's retention. Each recommendation includes severity, specific fix, and expected retention lift.

create_edit

Create an Atlas auto-edit job from a video URL. Returns a job ID immediately — the video is processed asynchronously. The agent should tell the user "I've started processing your video" and then poll with wait_for_edit or get_job_status.

wait_for_edit

Poll an edit job until it completes or fails. Checks every 5 seconds and returns the final result with the output video URL. Use this after create_edit to wait for the result.

create_edit_and_wait

Create an edit job AND wait for it to complete in one call. Combines create_edit + wait_for_edit for convenience.

get_job_status

Check the status of an editing job. Returns progress, ETA, human-readable status message, and when complete — the output video URL.

export_nle_timeline

Export an NLE-compatible timeline (FCPXML, EDL, or Premiere XML) from a completed edit. Import into Final Cut Pro, DaVinci Resolve, or Premiere Pro to refine AI cuts.

batch_score_videos

Score multiple videos in a single request. Useful for content teams evaluating a batch before publishing.

Environment Variables

| Variable | Required | Default | Description | |----------|----------|---------|-------------| | ATLAS_API_KEY | Yes | — | B2B API key | | ATLAS_API_BASE | No | https://autoeditor.app | API base URL |

Use Cases

• Media companies: Score videos before publishing to catch retention killers
• Ad agencies: Evaluate ad creative for viewer engagement before spending on distribution
• Movie studios: Score trailer cuts for audience retention probability
• Content teams: Batch-score a week's content queue in one call
• AI agents: Let Claude/OpenAI agents create edits, check status, and export timelines programmatically