fcp-mcp
The most capable MCP server for Final Cut Pro — 88 tools covering FCPXML editing, live FCP control, parametric puppets, and media analysis.
README
fcp-mcp — Final Cut Pro MCP Server
<!-- mcp-name: io.github.dreliq9/fcp-mcp -->
The most capable MCP server for Final Cut Pro — 88 tools covering FCPXML editing, live FCP control, parametric puppets, and media analysis. One server, every layer.
You: "Open the library, find flash frames in the hero timeline, fix them, and bounce to ProRes."
Claude → fcp_open_library → fcpxml_detect_flash_frames → fcpxml_fix_flash_frames
→ fcp_import_xml → fcp_share → compressor_encode
Result: fixed FCPXML + ProRes master, verified in-app
Where the competition stops at one layer, fcp-mcp covers three:
- FCPXML engine — parse, edit, QC, generate, cross-NLE export — 49 tools
- Live FCP control — AppleScript-backed library/playback/menu/share — 20 tools
- Media analysis — ffprobe-backed scene/loudness/frame tools — 10 tools
- Parametric puppets — character animation presets in FCPXML — 7 tools
- Compressor — dispatch encodes to Apple Compressor — 2 tools
Available Tools (88 across 12 categories)
| Category | Count | What it does |
|---|---|---|
| inspect | 8 | Parse, list clips/markers/effects/roles, analyze pacing, timeline stats, A/B diff |
| qc | 10 | Flash frames, gaps, duplicates, media links, frame rates, audio levels, safe zones, duration, schema, full QC report |
| edit | 12 | Markers, keywords, titles, audio, transitions, trim, split, delete, reorder, speed, role assign, reformat |
| heal | 3 | Fix flash frames, fill gaps, remove silence |
| batch | 4 | Markers, rename, role assign, apply transition across many clips |
| generate | 4 | New project/timeline, auto rough cut, montage from a shotlist |
| templates | 3 | List, apply, save FCPXML templates (reusable title/audio/role bundles) |
| io | 5 | Import SRT/EDL, export EDL + DaVinci Resolve XML + Premiere FCP7 XMEML |
| live | 20 | AppleScript-backed: library/events/projects, playback, menu/keyboard, share, discover effects & motion templates |
| puppet | 7 | Parametric character rigs in FCPXML with motion presets (walk, talk, wave, multi-scene composition) |
| media | 10 | ffprobe + ffmpeg: info, streams, EBU R128 loudness, silence, beat detection, scene detect, thumbnails, audio-to-MIDI |
| compressor | 2 | List Compressor settings, dispatch encode jobs |
Project Structure
fcp-mcp/
├── src/fcp_mcp/
│ ├── __init__.py
│ ├── __main__.py # python -m fcp_mcp.server
│ ├── server.py # FastMCP entry + all 88 tool handlers
│ ├── fcpxml/
│ │ ├── parser.py # FCPXML → Python object tree
│ │ ├── writer.py # object tree → FCPXML (lossless)
│ │ ├── models.py # TimeValue, Timecode, Clip, Timeline, etc.
│ │ ├── time_utils.py # rational-arithmetic timecode
│ │ ├── analysis.py # pacing, flash frames, gaps, duplicates
│ │ ├── validator.py # DTD-style structural validation
│ │ ├── generator.py # programmatic project/timeline creation
│ │ ├── puppet.py # character puppet system
│ │ └── diff.py # timeline A/B comparison
│ ├── fcp_control/ # AppleScript bridge for live FCP control
│ ├── media/
│ │ └── ffprobe.py # ffprobe wrapper for media analysis
│ ├── pipeline/ # multi-step workflows
│ └── utils/
│ ├── safe_xml.py # defusedxml hardening
│ └── paths.py # path resolution, expansion
├── tests/ # 107 tests
├── examples/
│ ├── quickstart.py # install verification
│ └── GALLERY.md # workflow gallery with prompts
├── pyproject.toml
├── server.json # MCP Registry manifest
├── smithery.yaml # Smithery directory config
├── LLM_GUIDE.md # operational guide for agents
├── WORKFLOWS.md # production recipes
├── CHANGELOG.md
├── ROADMAP.md
├── CONTRIBUTING.md
└── LICENSE
Setup
Prerequisites
- Python 3.10+
- macOS with Final Cut Pro (for live-control tools — FCPXML tools work anywhere)
- FFmpeg on
$PATH(formedia_*tools):brew install ffmpeg - Apple Compressor (optional, for
compressor_*tools)
Install
pipx install fcp-mcp
Or from source (for contributors):
git clone https://github.com/dreliq9/fcp-mcp.git
cd fcp-mcp
python3 -m venv .venv
source .venv/bin/activate
pip install -e ".[dev]"
Verify
source .venv/bin/activate
python -c "import fcp_mcp; print(fcp_mcp.__version__)"
pytest tests/ -v
Connect to Claude Code
claude mcp add-json fcp '{"type":"stdio","command":"fcp-mcp"}' --scope user
Or if you installed from source, point at your venv's Python:
claude mcp add-json fcp '{"type":"stdio","command":"/FULL/PATH/TO/.venv/bin/fcp-mcp"}' --scope user
Or edit ~/.claude.json directly:
{
"mcpServers": {
"fcp": {
"type": "stdio",
"command": "fcp-mcp"
}
}
}
Claude Desktop
Add the same config to your Claude Desktop config file:
- macOS:
~/Library/Application Support/Claude/claude_desktop_config.json
Verify connection
claude mcp list # from terminal
/mcp # inside Claude Code
Key Features
Three layers in one server
Most FCP MCPs pick a lane. DareDev256/fcpxml-mcp-server is FCPXML-only. elliotttate/finalcutpro-mcp is AppleScript/JXA-heavy with thin FCPXML. fcp-mcp does both — and adds media analysis (ffprobe) and a parametric puppet system on top. An agent can open a library, inspect the active timeline, patch the XML, re-import, trigger Share, and dispatch a Compressor job — end to end.
Rational-arithmetic timecode
Time values are stored as fractions ("720/24s") and only collapsed to
floats at display boundaries. Frame-accurate across 23.976 / 24 / 29.97 /
59.94 / drop-frame — no rounding drift when splitting, trimming, or
concatenating.
Hardened XML parsing
All .fcpxml reads go through defusedxml via utils/safe_xml.py —
XXE, billion-laughs, and external-entity attacks blocked by default.
Size and depth limits enforced. validator.py runs structural schema
checks before any destructive operation.
Cross-NLE export
Same timeline out to three targets:
fcpxml_export_resolve— DaVinci Resolve-flavored XML (v1.9)fcpxml_export_fcp7— Premiere-compatible XMEML (FCP7 format)fcpxml_export_edl— flat EDL for color-grading and archive pipelines
Parametric character puppets
puppet_* tools build animated character rigs entirely in FCPXML —
no third-party motion templates required. Parts, keyframes, and
presets (walk, talk, wave, multi-character compositions) emit
standards-compliant XML that opens in any FCP.
QC before you cut
fcpxml_qc_report("hero.fcpxml")
→ flash frames (2 × 1-frame orphans at 00:12:03, 00:14:22)
→ audio-level warnings (3 clips peaking above -3dBFS)
→ offline media (1 stock clip missing)
→ safe-zone violations (title at y=0.94, outside 90% action safe)
→ frame-rate conflicts (none)
One call, full catalog of issues with timecodes, before the edit leaves your machine.
Live FCP control
When FCP is running, fcp_* tools wire through AppleScript:
fcp_open_library(...)
fcp_get_timeline_info() → current project, active range, playhead
fcp_playback("play" | "pause" | "goto" | "in_out")
fcp_menu_command("File > Export > Export Using Compressor Settings...")
fcp_keyboard_shortcut("cmd+shift+e")
fcp_share("YouTube — 4K")
fcp_discover_effects() → every installed effect/transition/title
Output Files
By default, modified FCPXMLs are written next to their input with a
_modified suffix. Override per-call via the output_path parameter,
or globally with:
export FCP_MCP_OUTPUT_DIR=/your/path
Examples
See the full gallery for tool sequences and workflow breakdowns. Sample prompts:
- "Run a full QC report on hero.fcpxml and fix every flash frame."
- "Convert these .srt captions into FCPXML title clips on the V2 track."
- "Take my assembly-edit XML and re-export it as a DaVinci Resolve XML for color."
- "List every clip on the timeline, then batch-assign the 'dialogue' role to all interview clips."
- "Open the active library, seek to the first marker, and dispatch a ProRes 422 HQ bounce via Compressor."
- "Build a 3-character puppet scene: one walking, one talking, one waving. 5-second timeline."
Architecture
Claude Code / Claude Desktop / any MCP client
│
│ stdio (JSON-RPC)
▼
fcp-mcp server (fcp_mcp/server.py, FastMCP)
│
├── fcpxml/parser + writer + models ← rational-arithmetic, defusedxml-hardened
├── fcpxml/analysis + validator ← QC, pacing, flash frames, gaps
├── fcpxml/generator + puppet ← programmatic timeline construction
├── fcpxml/diff ← A/B timeline comparison
├── fcp_control/ (AppleScript) ← live FCP when available
├── media/ffprobe ← clip info, loudness, scenes, frames
└── Compressor (CLI dispatch) ← automated encodes
│
▼
.fcpxml (v1.11+) / EDL / DaVinci XML / FCP7 XMEML / ProRes / H.264
Troubleshooting
fcp-mcp: command not found — make sure the venv you installed into is on
your PATH, or use the absolute path to the venv's bin/fcp-mcp in your MCP config.
ffmpeg: command not found on media_ calls* — brew install ffmpeg. FFmpeg
is not bundled.
fcp_* tools return "Final Cut Pro is not running" — launch FCP first. Live
tools require an active FCP process on macOS.
fcp_* tools fail silently on newer FCP versions — FCP's scripting surface is
limited compared to pre-FCPX versions. Menu items and keyboard shortcuts are the
reliable path; some JXA queries are gated behind Accessibility permissions
(System Settings → Privacy & Security → Accessibility → Claude/Terminal).
fcpxml_* tools complain about schema — Make sure your input is FCPXML 1.10+.
Run fcpxml_validate first. For older exports, open and re-save the library in
FCP 10.6+ before processing.
Tests failing on import — activate the venv and reinstall: pip install -e ".[dev]".
Planned Work
See ROADMAP.md for the full plan. Highlights:
- MCP Prompts — shipped:
qc-check,youtube-chapters,cleanup,rough-cut,beat-sync - Windows/Linux parity for FCPXML tools — live-control tools remain macOS-only
- Proxy/Resolve round-trip — proxy-aware offline/online workflows
Acknowledgments
fcp-mcp was co-developed by Adam Steen and Claude (Anthropic).
License
MIT — see LICENSE.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.