resolve-mcp: Skill for Claude Code

/review-cut — Review Cut + AI Visual Feedback

Render a lightweight review export, then upload it to Gemini so the AI can actually watch the timeline and give feedback. Optionally save the review file for sending to clients.

Arguments

• No arguments: render review, upload to Gemini, get AI feedback
• save to /path/: also save the review render to disk
• for client: render with watermark text (e.g., "REVIEW COPY — NOT FOR DISTRIBUTION")
• feedback on pacing: give Gemini a specific focus area

Workflow

1. Get timeline info

Use resolve_get_timeline_info to get the timeline name, duration, frame rate, and resolution.

2. Render a low-res review cut

Set up a lightweight render specifically for AI review and client screening:

• Use resolve_set_render_format_and_codec → MP4 / H.264
• Use resolve_set_render_settings with:

  {
    "SelectAllFrames": true,
    "FormatWidth": "1280",
    "FormatHeight": "720",
    "VideoQuality": "10000000"
  }
  

  (720p, low bitrate — fast to render, small enough to upload to Gemini)

Burned-in timecode: Check if a burn-in preset exists with resolve_get_render_presets. If there's one with "burn" or "timecode" in the name, load it. Otherwise, set render settings to enable data burn-in if the API supports it, or inform the user to enable it manually in the Deliver page.

Watermark: If the user requested for client, note that the watermark text should be set manually in Resolve's burn-in settings (the scripting API has limited burn-in control). Mention this to the user.

• Use resolve_add_render_job to queue
• Use resolve_start_render to begin
• Poll with resolve_get_render_status until complete

3. Upload to Gemini for visual analysis

This is the key differentiator — the AI actually watches the edit.

• Use resolve_analyze_timeline which uploads the timeline's source proxies to Gemini and runs a full editorial critique
• This gives Gemini visual context of what's actually on the timeline