Ableton Mcp Lofifren
About
Fork of ahujasid/ableton-mcp adding a 33-personality style system, ~35 extra MCP tools, batch commands, and arrangement-view BETA
Details
- Author
- LofiFren
- GitHub stars
- 3
- Downloads
- 213
- Categories
- Other
Jump to
- 33 rule-based player personalities (Coltrane, Bill Evans, Dilla, Metro Boomin, etc.)
- ~35 additional MCP tools across six tiers
- Bootstrap a full session with one call (setup_session)
- Composite tools: create_clip_with_notes, duplicate_clip, create_track
- Musical helpers: chord progressions, scale runs, drum patterns, quantization
- Personality blending via add_blended_personality_solo
- Arrangement view support (BETA): locators, loop, timeline clip placement
Setting up with Highlight
This MCP is not yet compatible with Highlight’s one-click setup. However, you can still use it with Highlight by following these steps:
- Download and install Highlight from highlightai.com/download
- Navigate to the plugins tab and select "Add Custom Plugin"
-
Configure the plugin with the settings below
Plugin Name
Ableton Mcp LofifrenCommand (node, npx, python, etc.)Please refer to the README for specific instructions on how to obtain API keys or other required environment variables.
- Enable "Start Automatically" if you want the plugin to start when Highlight launches
From the repository
Install the package from PyPI with pip install ableton-mcp-lofifren or run it directly using uvx ableton-mcp-lofifren. Configure it as an MCP server in Claude Desktop, Cursor, or VS Code using the provided install buttons; then connect it to a running Ableton Live session.
Claude Desktop / Cursor
Paste into your MCP client config file to install this server.
{
"mcpServers": {
"ableton mcp lofifren": {
"ableton-mcp-lofifren": {
"command": "uvx",
"args": [
"ableton-mcp-lofifren"
]
}
}
}
}
McpServers
{
"ableton-mcp-lofifren": {
"command": "uvx",
"args": [
"ableton-mcp-lofifren"
]
}
}
AbletonMCP - Ableton Live Model Context Protocol Integration
> Fork notice: This is a fork of ahujasid/ableton-mcp extending it with a 33-personality style system and ~35 additional MCP tools. The original project is by Siddharth Ahuja and remains MIT-licensed. This fork is published to PyPI as ableton-mcp-lofifren.
AbletonMCP connects Ableton Live to Claude AI through the Model Context Protocol (MCP), allowing Claude to directly interact with and control Ableton Live. This integration enables prompt-assisted music production, track creation, and Live session manipulation.
Join the Community
Give feedback, get inspired, and build on top of the MCP: Discord. Made by Siddharth
What's New in This Fork
This fork extends the upstream ahujasid/ableton-mcp with significantly more capability — the goal is a full song in one prompt. The original PR #84 changes (clip editing, mixer controls, scene firing, audio-track creation, undo) are still present and proposed upstream as ahujasid/ableton-mcp#84. On top of that, this fork now adds ~35 additional MCP tools across six tiers, plus a 33-personality rule-based style system spanning jazz, fusion, R&B, hip-hop, and modern trap — from Coltrane and Bill Evans to Timbaland and Metro Boomin.
Tier 1 — Composite "song scaffold" tools (one call, many steps)
- setup_session(tempo, time_signature, tracks=[…]) — Bootstrap an entire session in a single round-trip: tempo, time signature, plus N tracks each with optional name, instrument, volume, and pan.
- create_clip_with_notes(track, clip, length, notes, name?) — Fuses create_clip + add_notes_to_clip + set_clip_name.
- create_track(type, name?, instrument_uri?, volume?, pan?, index?) — Single-call replacement for the create → name → load chain.
- duplicate_clip(src_track, src_slot, dst_track, dst_slot) — Cross-track clip duplication. (Upstream's duplicate_clip_to only works within a single track.)
Tier 2 — Missing primitives
- Time signature — set_time_signature(numerator, denominator)
- Clip editing — set_clip_loop, set_clip_length
- Mixer state — set_track_arm, set_track_mute, set_track_solo, set_master_volume
- Track lifecycle — delete_track
- Devices — set_device_parameter (raw Live values, clamped to the parameter's min/max)
- Scenes — create_scene, set_scene_name, set_scene_tempo
Tier 3 — Browser improvements
- search_browser(query, category?) — Walk Ableton's browser tree by name (case-insensitive). Returns up to 50 matches with URIs ready to load.
- load_instrument_by_name(track_index, name, category?) — Server-side composite: search + load in one call. No more URI hunting.
- get_track_devices(track_index) — List the devices on a track plus every device's parameters with their value, min, and max — needed before calling set_device_parameter.
Tier 4 — Musical helpers (pure server-side, no remote round-trips for note math)
- add_chord_progression(track, clip, chords, rhythm?, octave?) — Write a progression like ["Cm","Fm","G7","Cm"] straight into a clip. Supports triads, 7ths, sus, dim, aug, 9ths, 11ths, 13ths.
- add_scale_run(track, clip, scale, start_beat, end_beat, direction?) — Generate scalar runs in any of major / minor / harmonic_minor / dorian / phrygian / lydian / mixolydian / locrian / pentatonic / blues / chromatic.
- add_drum_pattern(track, clip, pattern, length, kit_map?) — Preset patterns: four_on_floor, trap, breakbeat, boom_bap. Defaults to GM drum mapping.
- quantize_clip(track, clip, grid?) — Snap notes to 1/4, 1/8, 1/16, or 1/32.
- transpose_clip(track, clip, semitones) — Shift every note by a fixed interval.
Tier 4 Plus — Personality system (27 named players, 4 roles, BPM-aware)
A personality is a rule-based profile capturing the surface stylistic features of a real-world player: note density, swing, register, scale pool, chromatic usage, dynamics, articulation. Generators combine those rules with a chord progression to produce a coherent MIDI part that reads as "in the style of" that player. Not ML — deterministic, rule-based, fast, extensible (a new personality is one dict entry).
Each personality declares a tempo_sweet_spot and a comfortable tempo_min/tempo_max range. Generators auto-pull the current session tempo and scale density / swing to compensate when the song tempo is outside the personality's range — and emit a warning so you know.
Tools
- add_personality(track, clip, personality, chord_progression, bars_per_chord?, tempo?, octave_offset?, seed?) — Generate a part in the named personality's style and write it into an existing MIDI clip. Auto-dispatches by role (solo / comp / bass / drums). Tempo defaults to the live session tempo.
- add_blended_personality_solo(track, clip, personality_a, personality_b, ratio, chord_progression, ...) — Generate a solo from a profile that's interpolated between two personalities (e.g. 70% Coltrane + 30% Kenny G). Numeric fields are linearly blended; categorical pools take from A when ratio<0.5, from B otherwise.
- list_personalities() — Returns every personality grouped by role with name, sweet-spot tempo, and one-line description.
- add_personality_solo(...) — Legacy alias for add_personality restricted to solo personalities.
Personality reference (33 total)
Solo (9) — melodic / lead lines
| Key | Player | Sweet spot | Surface style |
|---|---|---|---|
| coltrane | John Coltrane | 200 BPM | Sheets of sound, modal scalar runs with chromatic approach |
| kenny_g | Kenny G | 95 BPM | Sparse high-register sustained chord tones, lots of space |
| oscar_peterson | Oscar Peterson | 160 BPM | Swung bebop with octave doublings, blues turns, dramatic dynamics |
| miles_davis | Miles Davis | 120 BPM | Cool / modal — half the bar is silence, target chord tones |
| charlie_parker | Charlie Parker | 230 BPM | Bebop dense — chromatic approach into every chord tone |
| wayne_shorter | Wayne Shorter | 140 BPM | Modern angular — wide intervallic leaps, modal/altered |
| pat_metheny | Pat Metheny | 130 BPM | Lyrical guitar — long phrases over lydian/dorian, mostly straight |
| stan_getz | Stan Getz | 100 BPM | Cool / bossa — smooth long melodic phrases, mostly chord tones |
| dizzy_gillespie | Dizzy Gillespie | 240 BPM | Bebop trumpet — fast chromatic runs in the screaming register |
Comp (6) — chord voicings / comping
…
Sign in to leave a review
Use Google, GitHub, or an email account so ratings stay tied to real people.
No reviews posted yet.



