sfx
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| SFX_OUT_DIR | No | Where renders and sidecar specs go. Default: ./out | ./out |
| SFX_NO_PLAYBACK | No | Set to 1 to make sfx_play a no-op. | 0 |
| SFX_PRESETS_DIR | No | User preset folder. Default: %APPDATA%\sfx-api\presets | %APPDATA%\sfx-api\presets |
| SFX_NO_PEDALBOARD | No | Set to 1 to force NumPy effects even if pedalboard is installed. | 0 |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": false
} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
| experimental | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| sfx_docsA | Documentation for designing sounds. section: 'all' (compact schema + cookbook + preset names), 'schema' (compact field list), 'json_schema' (full JSON schema), 'cookbook', or 'presets' (full preset specs). Call with 'all' once before designing sounds. |
| sfx_list_presetsA | List available presets with a one-line summary of each. |
| sfx_renderA | Render a SoundSpec to a 16-bit mono WAV. Returns the file path, an id for later tweaks, features (active_ms, attack_ms, decay_ms, pitch_hz_start/end, band_db, lufs; definitions in the cookbook), normalization info, and warnings such as a missed loudness target. out_dir defaults to the SFX_OUT_DIR environment variable or ./out under the server's working directory. |
| sfx_render_presetB | Render a named preset, optionally with a JSON merge patch of overrides applied first. |
| sfx_tweakA | Apply a JSON merge patch to a previous render (by id or WAV path) and re-render. Example patch: {"layers": [...]} replaces all layers; {"master": {"target_lufs": -14}} changes one field. Lists are replaced wholesale, so send the full layers list when editing a layer. |
| sfx_variationsA | Render N randomized siblings of a sound (by id, path, or inline spec) for engine random containers. Jitter amounts come from spec.variation: pitch_cents, filter_cents, timing_ms, gain_db. Ids are -v1..vN. |
| sfx_playA | Play a WAV (by id or path) through the machine's default audio output so the user can hear it. |
| sfx_analyzeC | Measure any WAV: active_ms, attack_ms, decay_ms, pitch_hz_start/end, band_db, lufs, peak. Definitions in the cookbook. |
| sfx_save_presetC | Save a spec (inline, or from a render id) to the user preset library under a name. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
| schema_resource | JSON schema for SoundSpec |
| cookbook_resource | Sound-design recipes and feedback vocabulary |
| presets_resource | All presets, built-in and user-saved |
TDQS
Scored across 9 tools
Each tool has a clearly distinct role: docs, render, preset render, preset listing, tweak, variations, playback, analysis, and saving presets. The only similar pair—render and render_preset—is disambiguated by input type (inline spec vs. named preset).
Most tools follow a consistent sfx_<verb>[_<object>] pattern (render, render_preset, list_presets, save_preset, analyze, play). Minor deviations: sfx_docs and sfx_variations are noun-style rather than verb-based, so the pattern is not perfectly consistent.
Nine tools is well-scoped for a sound design toolkit. Each tool covers a distinct stage of the workflow without unnecessary redundancy or bloat.
The surface covers the full design workflow: document, render, tweak, randomize, listen, analyze, and save presets. Minor gaps like deleting or editing presets and managing previous renders are absent but not blocking for core sound design.