| name | aicw-video |
| description | Use AICW Video to turn local video/audio files into short clips, captions, rendered social videos, or step-by-step tutorials through the aicw-video MCP tools and local review hub. Trigger on requests involving local videos, screen recordings, screencasts, demos, shorts, clip planning, caption editing, tutorial export, or review links. |
AICW Video
AICW Video is a local CLI, MCP server, and browser review hub. Prefer the
aicw-video MCP tools and review UI over hand-written ffmpeg commands.
Default Workflow
- Call
create_project with the local files. Use autoMatchAudio: true unless the user asks otherwise.
- Call
analyze_project.
- Call
review_project and give the user the returned local URL.
- Let the user review ranges, captions, tutorial steps, privacy options, and render settings in the hub.
- If the user asks for edits, call
get_clip_plan, patch only the requested fields, then call save_clip_plan.
- Render only when the user asks: use
render_clip or render_all_clips.
Tutorials
For screencasts and demos, keep tutorials simple: key moments, screenshots, and editable text. The user can remove extra screenshots or text later.
- If the user asks for a tutorial from a selected clip, call
export_clip_tutorial.
- If no clip is selected, send the user to the review hub and point them to
Export tutorial.
Rules
- Preserve original source files.
- Do not re-run analysis if the project already has useful outputs unless the user asks.
- Do not manually edit generated mp4 files unless the user explicitly requests custom post-processing.
- If AI scene analysis is enabled, sampled frames/transcripts may be sent to the configured AI host. Local Whisper and local face detection do not require cloud AI.
- Return local paths and review URLs clearly.