clipcannon

name	clipcannon
description	Local-GPU video understanding and editing via MCP — analyse footage, find the best moments, cut highlight reels, add captions, render platform-ready clips (TikTok/Reels/ Shorts/YouTube), clone voices, lip-sync avatars, generate music, and run a real-time voice agent. Use when the user says "edit this video", "find the best moments", "create a highlight reel", "add captions", "clone voice", "lip sync", "render for TikTok", or "talk to Jarvis". Not for plain format conversion (ffmpeg-processing), image generation (comfyui), or meeting capture (echoloop).
version	0.1.0
author	ChrisRoyse
repo	https://github.com/JLMA-Agentic-Ai/jlma-clipcannon
mcp_server	true
protocol	stdio
entry_point	clipcannon serve
tags	["video","editing","voice-cloning","lip-sync","transcription","voice-agent","ai-music","text-to-video","mcp","gpu"]
env_vars	["CLIPCANNON_DATA_DIR","CLIPCANNON_GPU_DEVICE","CLIPCANNON_NVENC"]

ClipCannon -- AI Video Editor via MCP

Turns Claude into a professional video editor. Ingest video, run a 22-stage AI analysis DAG, then use 51 MCP tools across 12 categories to find moments, create edits, render platform-ready clips, generate music, clone voices, produce lip-synced talking-head videos, and converse via a real-time voice agent. Everything runs locally on GPU with a tamper-evident SHA-256 provenance chain.

Full catalog (51 tools, 14 ML models, 5 embedding spaces, credit costs, architecture diagram, Voice Agent lifecycle, integrations) lives in references/tools-and-models.md. Read it when you need an exact tool name, a model's VRAM budget, or the credit cost of an operation.

When to Use This Skill

• Video editing: "edit this video", "cut the boring parts", "create a highlight reel"
• Content discovery: "find the most emotional moments", "find where they talk about X"
• Platform rendering: "render for TikTok", "create Instagram Reels version"
• Voice: "clone this speaker's voice", "generate narration", "lip sync"
• Audio: "add background music", "generate sound effects", "compose a score"
• Analysis: "transcribe this video", "who are the speakers?", "scene breakdown"
• Text-to-video: "generate a video from this script" (end-to-end voice + lip-sync)
• Voice Agent: "talk to Jarvis", real-time conversational AI with wake-word activation

When Not to Use

• For simple video format conversion -- use ffmpeg-processing
• For AI image generation -- use comfyui or art
• For agentic video production from scratch -- use open-montage
• For meeting transcription -- use echoloop
• For audio-only processing -- use ffmpeg-processing

Quick Path

# Install (requires Python 3.12+, CUDA GPU, 8+ GB VRAM min, 24+ GB recommended)
pip install clipcannon
clipcannon serve                       # start the MCP server

# Docker (Dashboard :3200, License server :3100)
cd config && docker compose up -d

Typical MCP flow once the server is up: