ForgeFrame -- Workshop Video Brain
A local-first Claude Code plugin for tutorial video creators. Plan, edit, and publish workshop videos using AI-powered skills, 88 MCP tools, and a comprehensive video editing handbook.
New to video editing? The handbook is a complete guide from zero to published video -- covering Kdenlive, production concepts, and ForgeFrame automation.
Read the Handbook
Part
Chapters
What You'll Learn
Foundations
Ch.00-02
ForgeFrame setup, production pipeline, learning path
Preproduction
Ch.03-04
Outline, script, shot plan, capture checklist
Production
Ch.05
Camera settings, lighting, audio, VFR prevention
Postproduction
Ch.06-12
Kdenlive editing, transitions, color, audio, pacing, effects
Delivery
Ch.13-16
Export, QC automation, YouTube publishing, social clips
Reference
Ch.17-20
Troubleshooting, hardware, skill catalog, resources
Includes a workflow cheatsheet , Kdenlive shortcuts , and glossary .
Skill
What It Does
/ff-init
Initialize ForgeFrame environment with vault structure and config
/ff-new-project
Create a complete project (workspace + outline + script + shot plan) from a brain dump
/ff-video-idea-to-outline
Structured tutorial outline from a brain dump
/ff-tutorial-script
Full script with hook, materials, steps, mistakes, and conclusion
/ff-shot-plan
Production shot list covering 7 categories (A-roll, overhead, closeup, measurements, inserts, glamour, pickups)
/ff-capture-prep
Pre-shoot checklist: camera settings, audio, lighting, optimized shot order
/ff-broll-whisperer
B-roll suggestions from transcript analysis (5 shot type categories)
/ff-obsidian-video-note
Create/update a living Obsidian project note with section-safe sync
Skill
What It Does
/ff-auto-editor
Assemble a first-cut Kdenlive project from clips + script
/ff-rough-cut-review
Pacing and structure feedback from transcript + markers
/ff-pacing-meter
Per-segment energy analysis: WPM, speech density, energy drops
/ff-voiceover-fixer
Identify rambling sections with tightened rewrites
/ff-audio-cleanup
Full voice processing pipeline: denoise → EQ → compress → normalize → limit
/ff-pattern-brain
Extract materials, measurements, steps for MYOG overlays + printable notes
Skill
What It Does
/ff-publish
YouTube metadata bundle: titles, description, tags, chapters, summary
/ff-social-clips
Extract highlight clips for Shorts/Reels with platform-specific posts
/ff-youtube-analytics
Fetch and analyze YouTube channel data with engagement insights
Category
Tools
Count
Workspace
create, status, snapshot_list, snapshot_restore
4
Media
ingest, list_assets, proxy_generate, check_vfr, transcode_cfr
5
Transcript
generate, export
2
Markers
auto_generate, list
2
Clips
label, search, insert, remove, move, split, trim, ripple_delete, speed
9
Timeline
build_review, build_selects
2
Project
create_working_copy, validate, summary, setup_profile, match_source, archive
6
Transitions
apply, apply_at, apply_between
3
Compositing
composite_pip, composite_wipe
2
Effects
effect_add, effect_list_common
2
Color
color_analyze, color_apply_lut
2
Audio
fade, normalize, compress, denoise, enhance, enhance_all, analyze
7
Tracks
add, mute, visibility, gap_insert
4
Subtitles
generate, export
2
Render
preview, final, list_profiles, status
4
QC
qc_check
1
Assembly
plan, build
2
Analysis
broll_suggest, pacing_analyze, pattern_extract, replay_generate, title_cards_generate, voiceover_extract_segments
6
B-Roll Library
index, search, tag, stats
4
Social
find_clips, generate_package, clip_post
3
Publishing
bundle, description, titles, tags, summary, note
6
YouTube
fetch_channel, fetch_video, analyze, save_to_vault
4
ForgeFrame
init, status, project_new, project_list
4
Server
ping
1
As a Claude Code Plugin (recommended)
# Add the ForgeFrame marketplace
/plugin marketplace add Caleb68864/ForgeFrame
# Install Workshop Video Brain
/plugin install workshop-video-brain@forgeframe
This gives you all 17 skills and starts the MCP server automatically.
git clone https://github.com/Caleb68864/ForgeFrame.git
cd ForgeFrame
uv sync
Python 3.12+
uv package manager
FFmpeg (required for video processing)
faster-whisper (required for transcription)
Kdenlive (recommended for timeline editing)
/ff-new-project "I want to make a tutorial about building a dovetail joint"
This creates a workspace with outline, script, shot plan, and capture checklist in one shot.
/ff-capture-prep # Get your pre-shoot checklist
# ... film your tutorial ...
wvb media ingest ./workspace/ # Scan, proxy, transcribe, detect silence
/ff-auto-editor # Assemble a first cut
/ff-rough-cut-review # Get structural feedback
/ff-pacing-meter # Find slow sections
/ff-audio-cleanup # Polish the audio
wvb render final ./workspace/ --profile youtube-1080p
wvb qc-check ./renders/output.mp4 # Automated quality checks
/ff-publish # Generate YouTube metadata
/ff-social-clips # Extract clips for Shorts/Reels
Feature
What It Does
Transcript
faster-whisper local STT with SRT/JSON/text export
Auto-Marking
14 marker categories + phrase detection + repetition detection
Clip Labels
Auto-label clips by content (tutorial_step, talking_head, b_roll, etc.)
Review Timeline
Best-guess-first ranked review in Kdenlive
Selects Timeline
Filtered timeline of high-confidence segments
Replay Generator
Highlight cut with crossfades
Color Analysis
Color space detection, HDR flagging, LUT application
QC Automation
Black frame, silence, loudness, clipping, file size checks
VFR Detection
Scan for variable frame rate + auto-transcode to CFR
Render Profiles
YouTube 1080p/4K, Vimeo HQ, Master ProRes/DNxHR
Compositing
Picture-in-picture presets + wipe/dissolve transitions
Effects
Generic effect insertion + curated reference list
Project Archive
Streaming tar.gz/zip workspace bundling
Obsidian Sync
Section-safe note updates with frontmatter merge
Snapshots
Copy-first safety -- originals never touched
Layer
Technology
MCP server
FastMCP
Data models
Pydantic v2
CLI
Click
Templating
Jinja2
Video processing
FFmpeg / ffprobe
Transcription
faster-whisper
Build / deps
uv + hatchling
uv run pytest tests/ -v # 2,189 tests
MIT -- see LICENSE .