Skip to main content

Offline, deterministic, terminal-first MIDI DAW for agents

Project description

claw-daw

Offline, deterministic, terminal-first MIDI DAW.

Think “DAW for agents”: you (or an agent) write a small headless script (patterns + clips), render, listen, tweak, rerender — and keep everything diffable in Git.

What you can do

  • Headless scripting: build songs with new_project / add_track / new_pattern / add_note_pat / place_pattern / export_*
  • Deterministic output: same script + same seed + same version + same SoundFont → same render
  • Exports: WAV / MP3 / M4A / MIDI + project JSON
  • Sound engineering (mix spec, opt-in): deterministic per-track EQ/dynamics, sidechain, sends/returns, and audio metering
  • Quality workflow: claw-daw quality ... runs mix prep + preview/full gates + stem/bus checks
  • Agent ergonomics:
    • bar:beat timecodes (2:0, 1:3:120)
    • note expressions (chance, accent, mute, glide_ticks)
    • select/apply edits (select_notes, apply_selected)
    • validate/diff/analyze helpers
  • Sampler (offline): built-in drums + 808
    • Drum Kits v1 (role-based): list_drum_kits, set_drum_kit <track> <kit>, and add_note_pat ... <pitch|role> ...
    • 808 presets: set_808 <track> <preset> + set_glide for portamento
  • Genre Packs v1 (from-scratch, deterministic): claw-daw pack <trap|house|boom_bap> ...
  • Stylepacks v1 (opt-in / explicit request) (scored iteration + reports): claw-daw stylepack <trap_2020s|boom_bap|house> ...
  • Prompt → script helper (offline): style-aware scaffolding with novelty control (--max-similarity)
    • Supported styles: hiphop | lofi | house | techno | ambient | trap | boom_bap
  • MIDI out: play to hardware/virtual MIDI ports (claw-daw play)

Install

Linux (recommended)

curl -fsSL https://www.clawdaw.com/install.sh -o /tmp/clawdaw-install.sh
bash /tmp/clawdaw-install.sh

macOS

curl -fsSL https://www.clawdaw.com/install_mac.sh -o /tmp/clawdaw-install.sh
bash /tmp/clawdaw-install.sh

Windows (PowerShell)

iwr https://www.clawdaw.com/install_win.ps1 -UseBasicParsing -OutFile $env:TEMP\clawdaw-install.ps1
& $env:TEMP\clawdaw-install.ps1

Run in Administrator PowerShell to install system deps (Chocolatey).

Manual (Ubuntu):

sudo apt-get update
sudo apt-get install -y fluidsynth ffmpeg fluid-soundfont-gm pipx
pipx install claw-daw

If claw-daw is not found after install, run pipx ensurepath and restart your terminal.

Verify install

claw-daw --version
claw-daw doctor

Quickstart

1) Health check + SoundFont

claw-daw doctor
claw-daw paths --soundfont

2) Render a headless script (recommended workflow)

claw-daw --headless \
  --soundfont /usr/share/sounds/sf2/default-GM.sf2 \
  --script tools/my_song.txt

3) Prompt → script helper (optional)

This is a scaffolding tool. Best results come from iterating like a producer.

SF2=$(claw-daw paths --soundfont | head -n 1)
claw-daw --soundfont "$SF2" prompt \
  --out my_prompt_song \
  --prompt "modern trap, 140bpm, E minor, spacey" \
  --iters 8 --max-similarity 0.85 \
  --render --quality-preset edm_streaming --section-gain

# outputs:
# tools/my_prompt_song.txt
# out/my_prompt_song.mp3
# out/my_prompt_song.mid
# out/my_prompt_song.json
# out/my_prompt_song.meter.json
# out/my_prompt_song_stems/*.wav
# out/my_prompt_song_busses/*.wav

Omit --seed on prompt, pack, or stylepack to auto-pick a fresh seed for that run. Pass --seed N to replay the same result exactly.

Minimal script example (from scratch)

new_project minimal_demo 140
set_swing 16

add_track Drums 0
set_drum_kit 0 trap_hard

add_track 808 0
set_808 1 dist
set_glide 1 0:0:90

new_pattern 0 d1 2:0
add_note_pat 0 d1 kick 0:0 0:0:180 112
add_note_pat 0 d1 snare 0:2 0:0:180 108
place_pattern 0 d1 0:0 16

# (optional) a single bar bass pickup
new_pattern 1 b1 1:0
add_note_pat 1 b1 33 0:2 0:1:0 112
place_pattern 1 b1 0:0 16

save_project out/minimal_demo.json
export_midi out/minimal_demo.mid
export_mp3 out/minimal_demo.mp3 preset=clean
# macOS only: create a FL Studio project + sibling assets folder
# export_flp out/minimal_demo

MIDI out (hardware / virtual ports)

claw-daw midi-ports
claw-daw play out/minimal_demo.json --midi-out "YOUR PORT NAME"

Docs

  • docs/USER_GUIDE.md
  • docs/AGENT_COOKBOOK.md
  • docs/AGENT_PLAYBOOK.md
  • docs/ARCH_STYLEPACK_PIPELINE.md (prompt → BeatSpec → compile → render → score → iterate)

Notes / guarantees

  • 100% offline core (no network APIs).
  • Reproducibility depends on pinning: claw-daw version + SoundFont + seed/script.
  • Project format is JSON (current schema v7; migrations included).
  • FL Studio export writes a real .flp via the local FL Studio app on macOS and requires Accessibility access for osascript.

License

MIT

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

claw_daw-0.2.2.tar.gz (121.3 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

claw_daw-0.2.2-py3-none-any.whl (159.7 kB view details)

Uploaded Python 3

File details

Details for the file claw_daw-0.2.2.tar.gz.

File metadata

  • Download URL: claw_daw-0.2.2.tar.gz
  • Upload date:
  • Size: 121.3 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.7

File hashes

Hashes for claw_daw-0.2.2.tar.gz
Algorithm Hash digest
SHA256 855cbfbc059e2e4e68d26d977f775bf7cf0d65e055c99d3ef70b4089881828ba
MD5 587719414d18ba43a7229f5572b63b96
BLAKE2b-256 27e0bcd4bd65fbd56e4684ac1ec7331099616081bebb7a01a71e9dde01f4fc0a

See more details on using hashes here.

Provenance

The following attestation bundles were made for claw_daw-0.2.2.tar.gz:

Publisher: publish.yml on sdiaoune/claw-daw

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

File details

Details for the file claw_daw-0.2.2-py3-none-any.whl.

File metadata

  • Download URL: claw_daw-0.2.2-py3-none-any.whl
  • Upload date:
  • Size: 159.7 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.7

File hashes

Hashes for claw_daw-0.2.2-py3-none-any.whl
Algorithm Hash digest
SHA256 456fe764156283621ebc8fbf712ff7cc51a3c3c0fe9ef022f5899d8be0da755a
MD5 56d88b8c39e36b5279642633f8c80263
BLAKE2b-256 b12f32f37d43d9d18e6a5ffb6764c8176664b756d3ba1a5c95200dbf2b0ae016

See more details on using hashes here.

Provenance

The following attestation bundles were made for claw_daw-0.2.2-py3-none-any.whl:

Publisher: publish.yml on sdiaoune/claw-daw

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page