Skip to main content
VEA (Video Editing Agent) is an AI editing system for long-form media. It automates the path from understanding source footage to exporting short-form video and editing artifacts.

Long-Form Understanding

Uses Memories.ai-powered indexing and comprehension to reason over long videos.

Narrative Generation

Generates recap/highlight scripts from your prompt and project context.

Clip Planning + Rendering

Selects relevant clips, composes timeline, and renders final output.

Production Outputs

Exports both .mp4 deliverables and Final Cut Pro XML for downstream editing.

​
Pipeline

  1. Index source media to build searchable understanding artifacts.
  2. Generate structured response plan (recap, highlights, or custom narrative).
  3. Select and assemble clips based on semantic relevance.
  4. Apply narration/music/cropping/subtitles according to options.
  5. Export outputs for distribution and editor handoff.

​
API Workflow

Index first: 
curl -X POST http://localhost:8000/video-edit/v1/index \
  -H "Content-Type: application/json" \
  -d '{
    "blob_path": "data/videos/MyProject/",
    "start_fresh": true
  }'
Then generate: 
curl -X POST http://localhost:8000/video-edit/v1/flexible_respond \
  -H "Content-Type: application/json" \
  -d '{
    "blob_path": "data/videos/MyProject/",
    "prompt": "Create a 2-minute recap of this movie",
    "video_response": true,
    "music": true,
    "narration": true,
    "subtitles": true
  }'

​
Setup Essentials

git clone https://github.com/Memories-ai-labs/vea-open-source.git
cd vea-open-source
uv sync
cp config.example.json config.json
Required configuration values:
  • MEMORIES_API_KEY
  • GOOGLE_CLOUD_PROJECT
  • ELEVENLABS_API_KEY
Also required:
  • Python 3.11+
  • FFmpeg installed
  • gcloud auth application-default login before runtime

​
Output Structure

Generated artifacts are typically saved under data/outputs/{ProjectName}/:
  • Final rendered video (.mp4)
  • Final Cut Pro XML (.fcpxml)
  • Intermediate assets such as clip plan, narration files, and music assets

​
Best Fit

  • Movie/documentary recap generation
  • Highlight reel automation for media teams
  • Long-to-short content repurposing pipelines
  • Editor-assist workflows that still need timeline-level control
Repository: Memories-ai-labs/vea-open-source