Skip to main content
Glama

omnidaw-agent-mcp

An unofficial, policy-enforced Model Context Protocol (MCP) server for OMNi DAW.

Purpose

While OMNi DAW includes an official MCP server, omnidaw-agent-mcp is an independent integration that adds:

  • Strict Operator Policy: Enforces explicit authorization (--allow-write, --allow-risk) rather than relying purely on descriptive model annotations.

  • Workflow Previews: Two-phase composable operations (preview -> inspect -> apply) with preflight validation, symbolic references ($ref), and per-step execution journals. Execution is not an atomic transaction: earlier steps may have changed state when a later step fails or has an unknown outcome. There is no automatic rollback or replay; inspect live state before recovery, and do not assume undo can reverse file writes.

  • Domain Filtering: Selectively expose API domains at startup, reducing context overhead and constraining backend actions.

  • Job & File Safety: Safe async job supervision with deadlines, strict loopback TCP restrictions, automatic token redaction from errors and logs, and verification of non-silent PCM audio.

  • Honest Offline Status: Accurate reporting of whether API capabilities originate from live discovery or bundled references.

Related MCP server: ComposerBench

Clip placement, checkpoints, and MIDI import

  • Master effect inspection: bundled official 0.21.0 params_list supports a required track, not a master chain. No fabricated chain argument is added. devices_list({"chain":"master"}) and devices_get({"chain":"master","slot":0}) already support master effects; this is not equivalent to paginated plugin parameter discovery.

  • Repeated clip placement: agent_repeat_preview takes track, slot, native_length_beats, count, and optional start_beat. Supply the actual native length in quarter-note beats; it is not inferred from undocumented engine fields or audio duration/tempo. Each placement has explicit start and length, computed from the original decimal values rather than accumulated float addition. Fractional lengths are supported within official placement bounds (0.25..4096 beats). This places existing session clips without clearing the track. The bundled arrangement.place_clip description mentions MIDI only; bounded live verification also confirmed repeated audio placement with explicit lengths (see verification scope below).

  • Batched arrangements and checkpoints: agent_arrange_timeline and repetitions group placements plus markers in ordered batches of at most 32 mutations. Markers follow placements and count toward the bound. More than one batch requires checkpoint_paths: exactly one distinct, new absolute .omni destination between adjacent batches, inside --output-root. Saves use the official session.save API and change the project's save destination. A missing/domain-filtered save method blocks preview. Saves count toward the unchanged 128-step plan limit. Single-batch existing calls remain compatible. Preview/apply retains process-local 300-second expiry and single-use plans. Partial failures return completed indices, failed-or-unknown index, pending indices and the full step scope; writes stop immediately. No rollback, automatic retry or replay is provided, including unknown mutation/checkpoint outcomes. Apply progress notifications are best-effort: callback exceptions are recorded as progress_warnings, never relabel confirmed writes, interrupt job supervision, or trigger replay. Checkpoint preview enforces the .omni suffix; existing destinations are refused even with overwrite authorization because checkpoints never request overwrite. Apply rechecks destination availability before any plan writes (and again at each save).

  • Local MIDI import: agent_midi_preview({"path":"/authorized/synthetic.mid","track":0,"slot":0}) reads a bounded local SMF or RIFF RMID under the existing --output-root policy (also used for this local input). Absolute paths only; resolved symlink escapes are denied before reading. It previews a replacement clips.write_midi; applying still needs write and risk authorization. Formats 0/1 and PPQ are supported, with FIFO overlap pairing per source track/channel; tracks/channels merge into one destination with an explicit warning. Note times are ticks / PPQ, independent of tempo. Tempo, controller, program and meta payloads are preserved in the report, not applied; sustain/instrument behavior may differ. Format 2, SMPTE, SysEx/system, pressure/pitch bend, malformed structures and unrepresentable note timing are refused before plan creation. Limits: 8 MiB, 100,000 events, 256 tracks, 4,096 notes, 1,024 clip beats. Empty note files are rejected.

Live verification scope (2026-10-28, Linux, API 0.21.0, app 0.1.0) exercised the actual SDK stdio bridge with generated local fixtures. Four explicit two-beat audio placements rendered eight correctly timed pulse windows with continuous low-level beds. A 33-placement plan saved a checkpoint after 32 new placements; project JSON and live arrangement readback confirmed the content. Reopening remains unverified: no session.open method exists. SMF and RMID imports returned the expected notes and a native-instrument render was non-silent; format 2 was explicitly refused. These are bounded fixture tests, not subjective listening or long-session reliability certification.

Live discovery also confirmed master builtin parameter inspection through devices.get, EQ per-band Mid/Side processing (a Side low-cut causally suppressed a synthetic 60 Hz side component while preserving 1 kHz), Saturator algorithms/oversampling/mix, Filter Notch mode, and sampler fine tune/pitch-envelope parameters through params.list / params.set. Parameter keys are instrument-specific: discover them on the actual target track, never transfer numeric keys between instruments. General parameter automation with smooth curves was accepted and read back; exponential interpolation and audible automation behavior remain unverified. External compressor routing, track polarity inversion, per-band correlation/vectorscope and group-output routing were not established; absence of a catalog field is not proof of GUI absence.

Optional sample search and audio diagnostics

The base server needs neither component. Optional integrations can be installed via package extras ([search], [diagnostics], [all]). When installed, users only configure paths to their data, never to source checkouts:

# Enable sample search with data paths; diagnostics are discovered automatically when installed
omnidaw-agent --allow-write --output-root /authorized/new-project \
  --sample-index-db /authorized/index.sqlite --library-root /authorized/library

Optional tools are listed only when their components and required data paths exist; agent_status.optional_integrations reports availability and source (package or repo). Runtime dependency/DB health is checked on each call. No indexing, migrations, component installs, global reloads or audio uploads occur. No reinstall is needed between calls. Paths shown are examples, not defaults:

  • agent_resource_search: call with {"filters":{"ext":"wav","bpm_min":140,"bpm_provenance":"est"},"limit":5}. The tool schema lists supported filters, which mirror reports.query; unknown filters, numeric strings, nonfinite/inverted bounds and unsupported provenance are rejected. Limit is 1–50 over MCP. Known corruption is excluded by default; include_errors/only_errors explicitly opt in. Missing paths are never included. A bounded isolated subprocess calls the component library over SQLite URI mode=ro plus query_only. The component CLI is deliberately not used: its connection enables writable WAL even for query. Mixed-root indexes fail closed. Resolved current paths must be files inside the configured library. Indexed SHA/measurements/descriptors and uncertainty are retained; current contents are not rehashed and metadata freshness is not certified.

  • agent_export_and_diagnose_preview, then agent_apply: {"path":"/authorized/new-project/mix.wav","save_path":"/authorized/new-project/checkpoint.omni","start_beat":0,"end_beat":8,"bpm":120,"intents":["Intentional breakdown"]}. Inspect the returned changes before apply. Requires a new .omni checkpoint, explicit finite beat range, normal write/risk policy and root containment. Timeline preparation uses the existing export workflow; checkpoint follows preparation and precedes render. Save changes the project's save destination. Only a confirmed completed render job proceeds to actual integer-PCM/non-silence verification and the local analyzer. Analysis uses argv without a shell; the analyzer subprocess has a 90 s deadline and a 2 MiB limit per output stream. Local path resolution, PCM verification, hashing and schema validation are outside that subprocess deadline: this is not a 90 s end-to-end guarantee. Only regular files are accepted; POSIX nonblocking open and descriptor checks reject FIFO replacement races. Filesystem stalls on regular files remain possible, and the authorized root must be under operator control (hostile ancestor-directory replacement is not sandboxed). Configured Python invocation paths preserve virtual-environment symlinks. JSON is checked against the configured component's report schema, supported analyzer version 0.1.0, path and before/after SHA. No diagnostic report files are written under Music (or elsewhere): results are returned through MCP JSON. Level-2 heuristics are never automatic defects.

  • agent_diagnose_wav is read-only recovery/offline checking for an existing WAV inside output-root: {"path":"/authorized/new-project/mix.wav","bpm":120}. It verifies PCM and invokes only the analyzer, never exports or saves. It does not establish that an arbitrary existing file came from a particular host job.

Failure states retain the apply journal: export_failed_or_unknown stops before analysis, export_outcome_unknown rejects missing/malformed completed-job evidence, wav_rejected preserves a completed export but refuses analysis, and analysis_failed preserves artifact/job/SHA without replaying export. Inspect live jobs after uncertain outcomes. Retry diagnostics on the existing file only, never retry an unknown export automatically. Missing config/component/DB, corrupt or incompatible DB, root escapes, deadlines and malformed/oversized output fail explicitly. Component code is trusted operator configuration, not a sandbox.

Verification on 2026-10-28: actual MCP stdio search returned real indexed files; actual MCP offline diagnostics analyzed a generated WAV with matching unchanged SHA. The combined host export-and-diagnose workflow was also exercised through actual MCP on an authorized disposable synthetic project: a new checkpoint, bounded four-beat render, completed job, PCM verification and analyzer SHA check passed. These bounded checks do not establish subjective musical quality, long-session reliability or behavior on other platforms.

Advanced / Development source checkouts

When developing against local source checkouts instead of installed packages, pass --sample-indexer-repo and/or --audio-diag-repo (with optional --sample-indexer-python and --audio-diag-python overrides):

omnidaw-agent --allow-write --output-root /authorized/new-project \
  --sample-indexer-repo /path/to/sample-indexer-checkout \
  --sample-index-db /authorized/index.sqlite --library-root /authorized/library \
  --audio-diag-repo /path/to/audio-diag-checkout

Platform Compatibility & Requirements

Requirements:

  • Python 3.12+ (standard library + mcp SDK)

  • OMNi DAW (v0.1.0 Build 5de6b9a or newer, with API enabled via --api or GUI Settings)

Supported Platforms:

  • Linux: Any standard modern Linux distribution with Python 3.12+ (Ubuntu, Debian, Fedora, Arch, etc.). Fully compatible with Wayland (e.g. niri, GNOME, Sway) and X11. Runs with the official OMNi AppImage.

  • Windows: Windows 10/11 (64-bit). Fully supported out of the box (uses %APPDATA%/omni/api/endpoint.json).

  • macOS: Supported by omnidaw-agent-mcp (reads ~/Library/Application Support/omni/api/endpoint.json and supports loopback TCP sockets). Note: OMNi DAW itself currently publishes official binaries for Linux and Windows; macOS will work immediately once OMNi releases a macOS build or when connecting via network tunnel to a remote host.

Installation

Standard installation with pip:

# Base server
pip install omnidaw-agent-mcp

# With all optional integrations (sample search + audio diagnostics)
pip install 'omnidaw-agent-mcp[all]'

# Or with individual extras:
pip install 'omnidaw-agent-mcp[search]'
pip install 'omnidaw-agent-mcp[diagnostics]'

For development:

git clone https://github.com/adrian-cancio/omnidaw-agent-mcp.git
cd omnidaw-agent-mcp

python3 -m venv .venv
source .venv/bin/activate   # On Windows: .venv\Scripts\activate
pip install -e '.[all,dev]'

To run diagnostics:

omnidaw-agent --diagnose

Running the Server

Speak MCP over standard I/O:

# Read-only safe inspection (default)
omnidaw-agent

# Allow project edits within authorized root
omnidaw-agent --allow-write --allow-risk --output-root /home/user/Music/projects

# Also let session.save / export.render replace existing files (only inside --output-root)
omnidaw-agent --allow-write --allow-overwrite --output-root /home/user/Music/projects

# Restrict exposed tools to specific domains
omnidaw-agent --domains session,tracks,clips,theory

MCP Client Configuration

Claude Desktop / Hermes / OpenCode

Add the following to your MCP client configuration (opencode.jsonc, claude_desktop_config.json, etc.):

{
  "mcpServers": {
    "omnidaw": {
      "command": "/path/to/omnidaw-agent-mcp/.venv/bin/omnidaw-agent",
      "args": [
        "--allow-write",
        "--output-root",
        "/home/user/Music/omnidaw-exports"
      ]
    }
  }
}

Remote Agent Architecture Note

OMNi DAW's JSON-RPC API is strictly local (127.0.0.1) and authenticates via an ephemeral token generated per session. For agents operating remotely (e.g. on a separate VPS):

  1. Run omnidaw-agent locally on the workstation where OMNi runs.

  2. Establish a secure authenticated tunnel (e.g., WireGuard, Tailscale, or an authenticated mTLS reverse proxy) connecting the remote agent harness to a local transport bridge.

  3. Never expose 127.0.0.1 directly or persist/transmit the API token across unsecured networks.

Composable Workflows

  • agent_status: Inspect connection state, catalog provenance, and active operator policies.

  • agent_inspect: Return compact, LLM-optimized summaries of project state and tracks.

  • agent_preview: Generate an execution plan from official methods with preflight checks and symbolic references.

  • agent_apply: Execute an approved preview plan once, producing an execution journal and handling partial failures.

  • agent_loop_preview: Compose a deterministic four-part electronic loop (drums, bass, chords, melody) using live instruments and native theory generators.

  • agent_arrange_preview: Arrange scenes into sections, duplicate clips with transforms, and apply automation.

  • agent_arrange_timeline: Place a batch of clips (track, slot, start_beat, length_beats) and optional markers in one preview plan.

  • agent_measure: Measure master/track loudness, tonal balance, and clipping, proposing mastering chains without perceptual claims.

  • agent_export_preview: Preview mixes or stems to WAV files with bounded job tracking. An explicit finite end_beat greater than start_beat (default 0) is required; the workflow never guesses song length. Failed or unavailable arrangement inspection produces no plan. Session-mode preparation selects the arrangement using arrangement.play (which starts playback) then stops it; this does not place session clips on the timeline or guarantee non-silent output. Inspect the range/content and verify the written audio separately.

  • agent_verify_wav: Verify that exported audio contains valid integer PCM with matching sample rate and duration. Audio whose peak is below silence_floor_dbfs (default -80 dBFS) is rejected as silent or dither-only. The result includes peak_dbfs and rms_dbfs.

Known Limitations

These are limits of the current OMNi API and of this server's safety model, not guarantees about audible results.

  • No native MIDI file import or session open. The host API exposes audio.import, session.info and session.save, but no native .mid import or session.open. This server's agent_midi_preview parses supported local MIDI files and previews a replacement clips.write_midi operation; it does not preserve all MIDI playback behavior (see formats and limits above). Notes can also be written with clips.write_midi or the theory/clip generators.

  • Audio import rounds native lengths. The imported clip length is rounded (observed: a 32 beat loop became 32.25 beats). Place audio clips with explicit native-length placements.

  • Lengthening an audio timeline clip did not repeat its content in observed use. The catalog documents repetition for MIDI tracks only; for audio, place the clip once per repetition with explicit native-length placements.

  • Exports can be structurally valid but silent. If the arrangement playback mode is wrong, an offline render can contain only dither. agent_verify_wav rejects audio whose peak is below silence_floor_dbfs (default -80 dBFS). This is a floor check: it does not prove the audio is continuous, complete or musical.

  • Plans are process-local. Previews live in memory inside the server process, expire after 300 s, hold at most 128 steps, and the oldest of 32 pending plans is dropped. Split large timelines into several agent_arrange_timeline batches and apply each one promptly.

Audio-stream failure on Linux (regression note, October 2026)

Earlier OMNi builds could lose their audio output stream during playback (log lines alsa::poll() returned POLLERR, then Audio stream died - auto-restarting engine). After that, audio stayed silent, including offline renders, and in some sessions the host process exited. The cause was never established, and no deterministic trigger was found. An earlier claim that about 75 s of accumulated analysis.capture calls triggers it was contradicted by later runs and has been removed.

With OMNi build bbf1bd36, whose developer reports related fixes, no new stream failures were seen in a ~30 minute session of mostly offline editing and exports, in a later session with ordinary 2 s agent_measure captures, or in a short playback test. This server applies no capture throttling. This is an observed improvement, not a certified fix: no long continuous playback test has been run, and behaviour on other operating systems is untested.

Practical guidance for MCP clients:

  1. Verify exported audio with agent_verify_wav or equivalent.

  2. Save the project often. If the host dies, do not restart it automatically; report the failure and the evidence.

  3. Prefer offline export over long real-time capture for evaluation.

Verification & Testing

Run unit tests and SDK subprocess integration tests:

pytest -v

Execute disposable integration verification against a target project:

python scripts/run_integration.py /path/to/endpoint.json /path/to/output_dir

License & Attribution

  • Code adapted from official OMNi DAW reference examples: Copyright (c) 2026 Tomasz Głuc, under the MIT License.

  • Extensions, workflows, and policy engine: Copyright (c) 2026 omnidaw-agent-mcp contributors, under the MIT License.

  • OMNi DAW is a proprietary audio workstation developed by Tomasz Głuc. This project is unofficial and independent.

Available Tools

115 tools
agent_applyB
Destructive

Apply a previously returned preview once, subject to operator startup policy

Inputs: plan_id (required).

agent_arrange_previewC
Read-only

Preview sections, duplication/transforms, automation and save

Inputs: sections (required), save_path, automation, transforms.

agent_arrange_timelineB
Read-only

Preview batches of at most 32 timeline mutations with inter-batch checkpoints

Inputs: placements (required), markers, checkpoint_paths.

agent_export_previewC
Read-only

Preview render/stems over an explicit beat range with bounded job waiting; no guarantee of audible content

Inputs: end_beat (required), path, folder, start_beat.

agent_inspectC
Read-only

Compact project inspection; metadata is untrusted data

Inputs: limit.

agent_loop_previewC
Read-only

Preview deterministic native four-part electronic loop using live instruments

Inputs: bpm, seed.

agent_measureC

Measure master or track, propose mastering chain, no perceptual claims

Inputs: source, target, seconds.

agent_midi_previewB
Read-only

Parse bounded local SMF/RMID under output-root and preview MIDI notes; metadata is reported, never applied

Inputs: path (required), slot (required), track (required).

agent_previewC
Read-only

Preview composable official calls; symbolic references use {$ref: alias.id}

Inputs: steps (required), label.

agent_repeat_previewC
Read-only

Preview consecutive placements using explicit native clip length in beats

Inputs: slot (required), count (required), track (required), native_length_beats (required), +3 more.

agent_statusB
Read-only

Catalog provenance, capabilities and enforced operator policy

Inputs: no named parameters.

agent_verify_wavA
Read-only

Read actual output PCM; check duration/rate/channels and non-silence (peak below silence_floor_dbfs, default -80, is rejected as silent or dither-only)

Inputs: path (required), channels, sample_rate, expected_seconds, +1 more.

agent_waitC
Read-only

Bounded job waiting with progress, no restart on timeout

Inputs: job (required), deadline.

analysis_captureA

Record what the master or one track plays for a few seconds (default 8, 1-30) and analyse it in the background: peak, true peak, RMS, crest factor, clipping, DC; LUFS integrated / short-term / momentary and loudness range, short-term per second; stereo correlation, width and mono loss; tonal balance (sub, bass, low-mid, mid, high-mid, high, air) and a 1/3-octave spectrum with per-band width. Starts playback if stopped. Returns a job id at once: call jobs.wait with it. One capture at a time.

Inputs: source (required), seconds, start_transport.

analysis_compareA
Read-onlyIdempotent

Where two finished captures (e.g. kick and bass, vocal and guitars) compete: band by band levels and the bands where both carry energy within a few dB of each other, strongest first. The usual cure is an EQ cut on one of them there (devices.add EQ + devices.set_params).

Inputs: a (required), b (required).

analysis_fileA
Read-onlyIdempotent

Analyse an audio file (WAV, FLAC, MP3, OGG, AIFF) the same way as analysis.capture: a reference track to compare the mix with (analysis.compare), or an exported master to master.check. Reads at most the first 10 minutes. Returns a job: jobs.wait for it.

Inputs: path (required).

arrangement_build_songA

Lay a song out on the timeline from session scenes: each section plays one scene row for a number of bars (e.g. intro 8, verse 16, drop 16, outro 8), every MIDI track's clip in that row copied and repeated to fill it, a marker per section; TOMNIC patterns go on as their notes. Audio tracks are reported as skipped. Refuses if the range already holds clips unless clear_existing. One undo step. Then arrangement.play.

Inputs: sections (required), markers, start_beat, clear_existing.

arrangement_clearA
DestructiveIdempotent

Remove timeline clips starting inside a range (default: all), on one track or all. Undoable.

Inputs: track, end_beat, start_beat.

arrangement_copy_clipA

Copy one timeline clip to another beat, on the same track or another one (to_track; MIDI clips go to MIDI tracks). The copy is independent: edit or mute it on its own to make a variation. Undoable.

Inputs: clip (required), track (required), to_beat (required), to_track.

arrangement_edit_clipA
Idempotent

Change one timeline clip (index from arrangement.get): move it, change its length (MIDI grows or shrinks, audio only shrinks), mute it, rename it, set its gain. Fields left out stay. E.g. mute the bass in the second half of a drop after copying it there. Undoable.

Inputs: clip (required), track (required), name, muted, +3 more.

arrangement_getA
Read-onlyIdempotent

The song on the timeline: its length, whether the arrangement or the session is playing, the loop region, the markers (sections) and every track's clips with start and length in beats.

Inputs: no named parameters.

arrangement_place_clipA

Copy one session clip onto its track's timeline at start_beat; length_beats longer than the clip repeats it. MIDI tracks. Undoable.

Inputs: slot (required), track (required), start_beat (required), length_beats.

arrangement_playA
Idempotent

Play the timeline from a beat: every track follows the arrangement again (like Back to Arrangement), the playhead moves, playback starts.

Inputs: from_beat.

arrangement_remove_clipA
Destructive

Delete one timeline clip (index from arrangement.get; later clips on the track move down one index). Undoable.

Inputs: clip (required), track (required).

arrangement_set_loopA
Idempotent

Set and enable (or disable) the arrangement loop region in beats. Undoable.

Inputs: end_beat (required), start_beat (required), enabled.

audio_importA

Load an audio file (WAV, AIFF, FLAC, MP3, OGG) into a session slot of an audio track, as dropping it from the browser does; arrangement.place_clip then puts it on the timeline. Needs file access, granted in Omni's Settings (AI Create). Undoable.

Inputs: path (required), slot (required), track (required).

audio_split_stemsA

Split a WAV mix into vocals, drums, bass and other with Demucs (the model downloads on first use). Returns a job; its result has the four file paths, which audio.import can load. Needs file access (Settings, AI Create).

Inputs: path (required), folder (required).

audio_transcribeA
Read-onlyIdempotent

Audio to notes with Basic Pitch: a file (path) or the audio in a session slot (track + slot). mode poly (chords), melody (one line) or bass. Returns a job; its result has the notes in beats at the project tempo, ready for clips.write_midi, and the detected length. File paths need file access (Settings, AI Create).

Inputs: mode, path, slot, track, +3 more.

automation_clearA
DestructiveIdempotent

Remove the automation of one parameter on a track (the lane stays, empty). Undoable.

Inputs: param (required), track (required).

automation_listB
Read-onlyIdempotent

The automation lanes of a track on the arrangement timeline: what each drives (its key), how many segments and points, the automated span in beats.

Inputs: track (required).

automation_writeA
Idempotent

Automate a track's volume, pan or any parameter over the song: points at song beats with values in the parameter's own unit; the lane is created if needed and its automation replaced. E.g. a filter sweep through a break: {param: "instrument:16", points: [{beat: 96, value: 400}, {beat: 128, value: 8000}]}. Plays when the arrangement plays. Undoable.

Inputs: param (required), track (required), points (required), curve.

clips_clearA
DestructiveIdempotent

Empty a session clip slot (notes, audio, sequencer). Undoable.

Inputs: slot (required), track (required).

clips_duplicateA
Idempotent

Copy a session clip to another slot, on the same track or another one of the same kind (MIDI to MIDI, audio to audio). Overwrites the target. Undoable.

Inputs: slot (required), track (required), to_slot (required), to_track.

clips_generate_melodyA
Idempotent

Write a melody into a MIDI slot with the piano roll's melody generator: a style (stepwise_flow, minimal_cell, classical_arch, funk_groove, celtic_reel, bebop_line, arpeggio_loop, synth_arp, sequence_run, ambient_drift, tension_build, modal_motive), a key, bars, register and density. It sits on the chords of the slot's chord lane, or of chords_from (e.g. the pad slot written by clips.write_progression). Same seed, same melody. Undoable.

Inputs: key (required), slot (required), style (required), track (required), +5 more.

clips_get_notesA
Read-onlyIdempotent

Notes of a session MIDI clip (pitch, start and duration in beats, velocity) and its loop length.

Inputs: slot (required), track (required).

clips_launchA
Idempotent

Play a session clip, quantised like a click on its slot. Starts the transport unless start_transport is false.

Inputs: slot (required), track (required), start_transport.

clips_listB
Read-onlyIdempotent

The session clip slots (scene rows) of a track: empty or not, kind (midi/audio/sequencer), loop length in beats, note count, playing / queued.

Inputs: track (required).

clips_set_lengthA
Idempotent

Set the loop length of a session clip in beats (notes past the end stay but do not play). Undoable.

Inputs: slot (required), track (required), length_beats (required).

clips_stopA
Idempotent

Stop the session clip of one track, or of every track when no track is given.

Inputs: track.

clips_transformA

Edit every note of a MIDI clip at once: transpose (semitones), quantize starts to a grid in beats (0.25 = 1/16 in 4/4) with a strength, scale and offset velocities. Undoable.

Inputs: slot (required), track (required), quantize, transpose, +3 more.

clips_write_drumsA
Idempotent

Write a drum pattern into a session slot of a TOMNIC track (TOMNIC plays its own step pattern, not clip notes). Each lane is one voice (kick, snare, hihat, bass, tom, perc, cymbal, fx, or 0-7) with a step string: x = hit, X = accent, . = rest; '|' and spaces may mark bars. At rate 1/16 sixteen steps are one 4/4 bar, e.g. kick "x...x...x...x...", snare "....x.......x...". Lanes may differ in length (polymeters). Replaces the pattern, or with mode merge only the lanes given. Undoable.

Inputs: slot (required), lanes (required), track (required), mode, +1 more.

clips_write_midiA
Idempotent

Write notes into a session clip slot of a MIDI track, creating the clip if the slot is empty. mode 'replace' (default) leaves exactly these notes, 'merge' adds them. length_beats sets the loop (default: the notes rounded up to whole bars). Times are in beats from the clip start (1 beat = a quarter note; in 4/4 a bar is 4 beats). The clip loops; launch it with clips.launch. Undoable.

Inputs: slot (required), notes (required), track (required), mode, +1 more.

clips_write_progressionA
Idempotent

Render a chord progression (text, library or generated, see theory.progression) into a session slot of a MIDI track and set the clip's chord lane, so the user sees the chords in the piano roll. The loop length is the progression's. Replaces the clip's notes. Undoable.

Inputs: slot (required), track (required), progression (required).

clips_write_slicesA
Idempotent

Write which slices of a sliced audio clip play when (replacing its notes): steps with a slice index, start and length in beats, velocity, and pitch in semitones, so one slice can play a melody (a vocal chop, a stab) and many slices a rearranged break. Undoable.

Inputs: slot (required), steps (required), track (required), length_beats.

devices_addA

Insert a builtin effect (a name from devices.types) into a chain, at the end or at 'position'. It starts at its default settings; shape it with devices.set_params. Returns the chain. Undoable.

Inputs: chain (required), device (required), position.

devices_getB
Read-onlyIdempotent

Every parameter of a builtin effect with its current value, range, default, unit (dB, Hz, ms, %) and group; a choice lists its options (index = value, e.g. Shape 4 = HighPass). Values are in those units, not 0..1.

Inputs: slot (required), chain (required).

devices_listC
Read-onlyIdempotent

The effects of a chain in signal order: a track's inserts ({"track": id}), a return bus ({"return_bus": n}) or "master". Each with slot, name, kind (builtin/plugin/script), bypass and parameter count.

Inputs: chain (required).

devices_removeA
Destructive

Remove an effect from a chain; later slots move up. Undoable.

Inputs: slot (required), chain (required).

devices_set_bypassA
Idempotent

Bypass (true) or enable an effect; the switch crossfades, no click. Useful to compare before / after with mix.meters. Undoable.

Inputs: slot (required), chain (required), bypassed (required).

devices_set_paramsA
Idempotent

Set several parameters of a builtin effect at once, by name (as devices.get lists them, case-insensitive) or index, in the parameter's own unit. All are checked against their ranges before any changes. One undo step. Returns the device with its new values.

Inputs: slot (required), chain (required), values (required).

devices_typesA
Read-onlyIdempotent

Omni's builtin effects that devices.add can insert, with their family: EQ, Filter, Compressor, Limiter, MultibandDynamics, Noise Gate, Transient Shaper, Saturator, Delay, Reverb, Chorus, Phaser and creative ones.

Inputs: no named parameters.

edit_redoA

Redo the step undone last. Returns what was redone.

Inputs: no named parameters.

edit_undoA

Undo the newest step in Omni's history (the user's or yours; every edit call of this API is one step labelled 'API: ...'). Returns what was undone.

Inputs: no named parameters.

export_renderA

Render the arrangement (start_beat to end_beat, default the whole song) to a WAV file, as the Export dialog does: in real time through the audio device, or with fast offline (faster than real time, the same mix), optionally normalised to a loudness target. Returns a job: jobs.wait until done; the result has the path and an analysis of the written file. Writes only absolute .wav paths and does not overwrite unless asked. Session clips play only if recorded into the arrangement.

Inputs: path (required), fast, dither, target, +4 more.

export_stemsA

Render one WAV per track (all audible tracks, or the ones listed) into a folder, offline and faster than real time: each stem is its track soloed, with its sends' returns, without the master chain unless through_master. Files are named ' .wav'. Returns a job whose result lists the files.

Inputs: folder (required), dither, tracks, end_beat, +4 more.

groove_getA
Read-onlyIdempotent

The project groove every clip follows (unless its track opts out): style, amount, grid, accent, lean, humanize; and which tracks play straight.

Inputs: no named parameters.

groove_setA
Idempotent

Change the project groove (fields left out stay): style (straight, mpc, jazz, nested, push, drunk, samba, waltz, custom), amount 0-1, grid 8th/16th/32nd, accent, lean_ms, humanize_ms, humanize_vel, seed. Picking a style without amount or grid takes the style's own. Undoable.

Inputs: grid, seed, style, accent, +4 more.

groove_set_trackA
Idempotent

Whether a track follows the project groove (false = it plays straight, e.g. a tight bass under swung hats). Undoable.

Inputs: track (required), follow (required).

instruments_listA
Read-onlyIdempotent

Instruments that tracks.create can load: name, family and the URI to pass as 'instrument'; installed plugins (group Plugin, with vendor and format) after the builtins. Families: Synth (Omni Synth = subtractive/wavetable all-rounder), Keys (Piano), Physical model (strings, winds, plucked, mallets), Drums (TOMNIC drum synth), Image to sound, DJ.

Inputs: no named parameters.

instruments_load_presetA
Idempotent

Load a preset file (from instruments.presets) into the plugin on a track: its whole sound, as the plugin's own preset browser would. Then params.list shows the new values to tweak. Undoable.

Inputs: path (required), track (required).

instruments_presetsA
Read-onlyIdempotent

Preset files (.fxp) installed with the external plugin on a track, e.g. Surge XT's factory patches: name, category (their folders) and path. Filter with query words ("bass reese", "leads"); pages of up to 200.

Inputs: track (required), limit, query, offset.

jobs_getA
Read-onlyIdempotent

State (running / done / failed), progress and, once done, the result of a job started by another call.

Inputs: job (required).

jobs_waitA
Read-onlyIdempotent

Wait until a job finishes (at most timeout_seconds, default and limit 9) and return it as jobs.get does; call again while it is still running.

Inputs: job (required), timeout_seconds.

markers_addA

Add a marker (or, with end_beat, a section region) to the timeline. Undoable.

Inputs: beat (required), name (required), color, end_beat.

markers_removeA
Destructive

Remove a marker by its index from arrangement.get. Undoable.

Inputs: index (required).

master_build_chainA

Put a mastering chain at the end of the master: an optional glue Compressor (2:1, slow attack, soft knee) and a true-peak Limiter with its ceiling at the target's. With a capture job of the master, the compressor threshold and the limiter's input gain are set from it to approach the target loudness. Refuses if the master already has a Limiter (adjust that one with devices.set_params). One undo step. Then capture again and master.check.

Inputs: target (required), job, glue.

master_checkA
Read-onlyIdempotent

Pass / fail of a finished master capture (analysis.capture of "master", or analysis.file of an exported mix) against a target: loudness gap in LU, true peak against the ceiling, the gain that would hit the target, and notes (short capture, clipping, over-limiting, phase).

Inputs: job (required), target (required).

master_targetsA
Read-onlyIdempotent

Delivery targets: streaming platforms (Spotify, Apple Music, YouTube, Tidal, Amazon Music, Deezer), podcast, broadcast (EBU R128, ATSC A/85) and a club convention, each with integrated LUFS, tolerance and true-peak ceiling.

Inputs: no named parameters.

midi_fx_addA

Append a MIDI effect from midi_fx.types to a track's MIDI FX chain, with its default settings. Undoable.

Inputs: track (required), device (required).

midi_fx_getB
Read-onlyIdempotent

Every parameter of one MIDI FX slot: name, value, range, default.

Inputs: slot (required), track (required).

midi_fx_listB
Read-onlyIdempotent

The MIDI FX chain of a track, in processing order.

Inputs: track (required).

midi_fx_removeA
Destructive

Remove one MIDI FX slot; later slots move down. Undoable.

Inputs: slot (required), track (required).

midi_fx_renderA
Idempotent

Play one MIDI FX slot offline over some bars and write what it played into a session slot as notes: a Markov / L-System / Marbles line becomes an editable clip, an Arpeggiator over the chord in input_slot becomes the arpeggio. The effect is bypassed afterwards (bypass false keeps it on). Undoable.

Inputs: slot (required), track (required), to_slot (required), bars, +2 more.

midi_fx_set_bypassB
Idempotent

Bypass (or re-enable) one MIDI FX slot. Undoable.

Inputs: slot (required), track (required), bypassed (required).

midi_fx_set_paramsA
Idempotent

Set parameters of one MIDI FX slot by name (or index), values in each parameter's range from midi_fx.get. Undoable, one step per call.

Inputs: slot (required), track (required), values (required).

midi_fx_typesA
Read-onlyIdempotent

MIDI effects a track can put in front of its instrument: transforms (Arpeggiator, Chord, Scale, Velocity, Note Length, Note Echo, Random, Ornaments, Tuplets) change the notes going through; generators (Marbles, Markov, L-System, Cellular, Thue-Morse, Recaman, Sequence, Shepard, Walks) play their own, locked to the song while a clip plays.

Inputs: no named parameters.

mixer_add_returnA

Add a return bus (e.g. 'Reverb', 'Delay'): put an effect on it with devices.add ({"return_bus": n}, usually at 100% mix) and feed it from tracks with mixer.set_send. Returns the mixer. Undoable.

Inputs: name.

mixer_getA
Read-onlyIdempotent

Master fader (dB) and balance, and the return buses (index, name, fader, insert count). Track faders and sends are in tracks.list.

Inputs: no named parameters.

mixer_remove_returnA
Destructive

Delete a return bus with its effects and every track's send to it. Later buses move down one index. Undoable (the sends come back too).

Inputs: bus (required).

mixer_set_masterA
Idempotent

Master fader in dB (up to about +6) and balance (-1..1). Prefer leaving the master at 0 dB and balancing the tracks. Undoable.

Inputs: pan, volume_db.

mixer_set_returnC
Idempotent

Fader of a return bus in dB. Undoable.

Inputs: bus (required), volume_db (required).

mixer_set_sendA
Idempotent

Send level from a track to a return bus in dB (up to +6; -120 or lower = off), e.g. to put a part into the reverb on a return. Undoable.

Inputs: bus (required), track (required), level_db (required).

mix_metersA
Read-onlyIdempotent

Live levels, as the mixer meters show them: per track peak/RMS (dBFS) and EBU R128 loudness (momentary, short-term, integrated LUFS, true peak dBTP); on the master also stereo correlation and loudness range (LU). Meters only move while audio plays: start the transport, wait a few seconds, then read. null = silence or not measured yet.

Inputs: tracks.

params_listA
Read-onlyIdempotent

Every parameter of a track: its instrument (builtin synth or VST3 / CLAP plugin) and every insert effect, plugins included, with key, name, device, group, range, default and current value. Plugins have hundreds: narrow with filter (words that must all appear, e.g. 'filter cutoff') and page with offset / limit (default 100).

Inputs: track (required), limit, filter, offset.

params_setA
Idempotent

Set parameters of a track by their params.list keys (instrument:, fx::), each in its own range (plugins are usually 0-1). All are checked before any changes; one undo step. For builtin effects devices.set_params by name is simpler.

Inputs: track (required), values (required).

sampler_getB
Read-onlyIdempotent

The sampler of an audio clip: file, length, playback mode, root note, transpose, envelope and its slices (index, the note that plays it, start and end in seconds).

Inputs: slot (required), track (required).

sampler_setA
Idempotent

Change how an audio clip's sampler plays (fields left out stay): playback_mode one_shot, loop, slice, gate, granular, warp; root note, transpose in semitones, reverse, envelope (attack/decay/release ms, sustain 0-1). Undoable.

Inputs: slot (required), track (required), reverse, sustain, +6 more.

sampler_sliceA
Idempotent

Cut an audio clip into slices, as the sample editor's slice detection does: on transients (sensitivity), into equal parts (count) or on beats (division per beat at the clip's tempo). A clip holds at most 92 slices (notes 36-127): slice a short loop, not a whole song. Switches the clip to slice mode: each slice gets a note, so clips.write_slices can play patterns and melodies from them. Returns the slices. Undoable.

Inputs: slot (required), track (required), mode, count, +2 more.

scenes_launchA
Idempotent

Launch every clip of a scene row together, quantised like a click on the scene. Starts the transport unless start_transport is false.

Inputs: scene (required), start_transport.

scenes_listA
Read-onlyIdempotent

Scene rows of the session: index, name, colour and how many tracks have a clip in the row. Clip slot N of every track belongs to scene N.

Inputs: no named parameters.

scenes_updateA
Idempotent

Rename or recolour a scene row (e.g. 'Intro', 'Drop'). Undoable.

Inputs: scene (required), name, color.

session_infoA
Read-onlyIdempotent

Overview of the open project: name, unsaved changes, sample rate, number of tracks, scenes and return buses, and whether the audio engine runs. Call this first to orient yourself.

Inputs: no named parameters.

session_saveA
Idempotent

Save the project with its audio collected next to it, to its own file or to an absolute .omni path (a new name becomes the project's). Does not overwrite another existing file unless asked.

Inputs: path, overwrite.

tempo_getB
Read-onlyIdempotent

The project tempo, the tempo map (tempo changes and ramps over the song) and the time signature changes.

Inputs: no named parameters.

tempo_set_mapA
Idempotent

Replace the tempo map: points at song beats with a tempo, each a step or a linear ramp to the next. Timeline clips, automation, markers and the loop stay on their beats. Empty points = one tempo (transport.set_tempo). A step at a section boundary plays exactly; a ramp inside a MIDI clip is followed block by block. Undoable.

Inputs: points (required).

tempo_set_time_signaturesA
Idempotent

Replace the time signature changes: from a bar on, so many beats of a unit (7/8, 3/4, …). Empty = the project signature everywhere. Undoable.

Inputs: points (required).

theory_analyzeA
Read-onlyIdempotent

The chords a MIDI clip's notes spell, with symbols and numerals, one per resolution_beats (default a bar); the key is estimated from the notes unless given. Use it to understand a user's part before writing around it.

Inputs: slot (required), track (required), key, resolution_beats.

theory_chordA
Read-onlyIdempotent

Notes of a chord symbol (C, Am7, F#m7b5, Bbmaj9, G7/B, Dsus4, …): pitch classes, names and MIDI notes in close position from the root.

Inputs: symbol (required), octave.

theory_harmonizeA
Idempotent

Find chords for the melody in a MIDI slot, in a key and a generator style (theory.vocabulary): one chord per harmonic-rhythm slot, chosen so the melody's strong notes are chord tones. fit 0 follows the style's habits, 1 the melody. Returns the chords; with write_to also writes them into a slot (performed like clips.write_progression). Undoable.

Inputs: key (required), slot (required), track (required), fit, +4 more.

theory_libraryA
Read-onlyIdempotent

Search the 240 named progressions (pop, rock, jazz, soul, blues, gospel, classical, modal, electronic, film, world, dark…). Pass a name to theory.progression or clips.write_progression as 'library'.

Inputs: limit, query.

theory_progressionA
Read-onlyIdempotent

Turn a chord progression into voiced, voice-led notes ready for clips.write_midi: from text (numerals or symbols), from the library, or generated in a style. Optional bass line and playing pattern (arpeggios, pulses, strums, …). Deterministic. To write it straight into a clip with its chord lane, use clips.write_progression.

Inputs: key (required), chords, library, perform, +1 more.

theory_scaleA
Read-onlyIdempotent

The seven degrees of a key (names, pitch classes, MIDI notes in the middle-C octave) and its diatonic chords as numerals, triads and sevenths. Use it to stay in key when writing melodies and bass lines.

Inputs: root (required), mode.

theory_vocabularyA
Read-onlyIdempotent

Every name the theory methods accept: modes, generator styles, playing patterns, library categories.

Inputs: no named parameters.

tracks_createA

Add a track at the end. kind 'midi' plays notes through an instrument: pass an instrument from instruments.list: a builtin URI or an installed plugin (its uri = path, or its name). Without one the track is silent. Plugins load in the background: their parameters show in params.list a moment later. kind 'audio' plays samples. Returns the new track. Undoable.

Inputs: kind (required), name, instrument.

tracks_deleteA
Destructive

Delete a track with its clips, devices and automation. Groups are not deleted this way (ungroup first). Indices of later tracks shift down. Undoable: undo rebuilds the whole track, instrument state included.

Inputs: track (required).

tracks_duplicateA

Copy a track (instrument, clips, effects, mixer settings, automation) right after it, named ' (Copy)'. A plugin instrument loads in the background: the copy shows in tracks.list a moment later. Undoable.

Inputs: track (required).

tracks_listA
Read-onlyIdempotent

Every track in mixer order with its id, kind (midi/audio/group/vca), instrument, fader (dB), pan, mute/solo/arm, colour, group, insert effects and send levels. Use the ids in later calls; indices shift when tracks move.

Inputs: no named parameters.

tracks_moveA
Idempotent

Move a track (a group with all its members) right before another track, joining that track's group, or to the end when 'before' is absent. Returns the new track order. Undoable.

Inputs: track (required), before.

tracks_set_instrumentA
Idempotent

Replace the instrument of a MIDI track with one from instruments.list (builtin, or an installed plugin, loaded in the background). The track keeps its name, clips, effects and mixer settings. Undoable: undo brings the previous instrument back with its sound.

Inputs: track (required), instrument (required), name.

tracks_updateA
Idempotent

Change a track's name, fader (volume_db, up to +6 dB), pan (-1..1), mute, solo or colour. Only the fields given change. Returns the track. One undo step (solo is monitoring and is not recorded).

Inputs: track (required), pan, mute, name, +3 more.

transport_getA
Read-onlyIdempotent

Play / record state, tempo (BPM), time signature and the playhead position in beats.

Inputs: no named parameters.

transport_playA
Idempotent

Start playback from the playhead. Does nothing if already playing.

Inputs: no named parameters.

transport_seekA
Idempotent

Move the playhead to a song position in beats (0 = song start).

Inputs: beat (required).

transport_set_tempoA
Idempotent

Set the project tempo in BPM (40-240). Arrangement clips keep their musical position. Undoable.

Inputs: bpm (required).

transport_stopA
Idempotent

Stop playback (and a running recording, which is then kept).

Inputs: no named parameters.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    B3.1/5.0

    Scored across 115 tools

    Disambiguation3/5

    The core DAW tools mostly target distinct resources and actions, but there is meaningful overlap between transport_get and tempo_get, transport_set_tempo and tempo_set_map, and jobs_wait versus agent_wait. The large agent_* family (agent_preview, agent_apply, agent_arrange_preview, agent_arrange_timeline, agent_measure, agent_export_preview) also duplicates native arrangement, mastering, analysis and export operations, making boundaries unclear. An agent can often tell tools apart, but the parallel native/preview/agent layers create real misselection risk.

    Naming Consistency4/5

    Most tools follow a predictable domain_action snake_case pattern (tracks_list, clips_write_midi, mixer_set_send, devices_add), which is easy to scan. Minor deviations exist, such as agent_* tools using a different naming style and some domain_verb forms like transport_play versus noun-like session_info, but the overall convention is stable.

    Tool Count1/5

    115 tools is far beyond a well-scoped surface, even for a DAW with many domains. The large agent_* preview/wrapper family adds substantial duplication on top of already complete native tools, making the set unwieldy and hard to navigate.

    Completeness4/5

    The surface covers an unusually broad DAW lifecycle: transport, tracks, session clips, arrangement, automation, devices, MIDI FX, mixer, theory, audio analysis, mastering, export and jobs. Gaps are minor, such as no explicit scene delete, limited note-level editing, and no track group/ungroup operation, but most workflows can be completed or worked around.

    Maintenance

    ActivityMaintained
    ResponsivenessNo recent community issues

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      B
      quality
      A
      maintenance
      Enables AI agents to control a browser-based digital audio workstation (openDAW) for music production, including track creation, effects, MIDI, automation, and rendering.
      100
      Apache 2.0
    • F
      license
      Not graded
      quality
      B
      maintenance
      Provides AI agents with reversible DAW control for music production, enabling inspection of instruments and parameters, batch edits, and evaluated revision runs with audio feedback and rollback.
      2
      -
    • A
      license
      B
      quality
      B
      maintenance
      Control Logic Pro locally on macOS through MCP: transport, tracks, mixer, plug-in parameters, audio analysis, bounce and delivery, with optional gesture and voice control. Uses OSC, CoreMIDI and macOS Accessibility. Actions report confirmed, uncertain or failed outcomes. Actual control requires a local Mac with Logic Pro. Hosted/Linux builds are schema inspection only; all operations are disabled.
      12
      3
      MIT
    • A
      license
      Not graded
      quality
      B
      maintenance
      Enables AI agents and scripts to edit, convert, repair, separate, tune, transcribe, notate, mix, and master audio through stable versioned JSON commands, with provenance sidecars and replayable, reproducible results. Every capability is exposed as tooling an agent can chain without a GUI, including dry runs, schema introspection, and pass/fail quality gates.
      5
      Apache 2.0