task-recovery
Recover the result of a timed-out Sonilo generation call using its task_id. Use whenever text_to_sfx, video_to_sfx, video_to_video_music, video_to_video_sfx, video_to_sound, video_to_video_sound, audio_ducking, dubbing, analyze_video, or video_to_music(preserve_speech=true) times out or its call was interrupted — the generation already ran (and was already charged) and its result is still retrievable.
sonilo-ai · AI review on file
Installation
Read the source and dependencies before running a command. Copying does not install or execute anything.
npx skills add sonilo-ai/skills --skill task-recoveryAbout this skill
name: task-recovery
description: Recover the result of a timed-out Sonilo generation call using its task_id. Use whenever text_to_sfx, video_to_sfx, video_to_video_music, video_to_video_sfx, video_to_sound, video_to_video_sound, audio_ducking, dubbing, analyze_video, or video_to_music(preserve_speech=true) times out or its call was interrupted — the generation already ran (and was already charged) and its result is still retrievable.
license: MIT
compatibility: "Requires Sonilo through either transport — the MCP server connected, or the sonilo CLI installed and signed in — plus credentials: a sonilo login sign-in, the hosted OAuth plugin, or SONILO_API_KEY. See the setup-api-key skill."
allowed-tools: Bash, Read, Write, mcp__sonilo__*
Sonilo Task Recovery
Every Sonilo generation that runs asynchronously on the backend (SFX, ducking, video-to-video, video-to-sound, dubbing, video-analysis, and video_to_music(preserve_speech=true)) is already charged the moment it's accepted — a client-side timeout does not stop it or refund it. This tool checks a task's status and, once finished, downloads its result. It never itself charges anything.
Setup: See the setup-api-key skill.
This tool is a safety net, not something to call speculatively. Only use it when you actually have a
task_id— from a timeout error message, or from the[sonilo-mcp] task submitted: <id>line a host prints to stderr the moment a task is submitted (this survives even a cancelled call).
Transport: MCP or CLI
Pick one at the start of the session and stay on it. Do not mix the two inside a single job, and do not announce the choice.
- Sonilo MCP tools visible in this session (
get_sfx_taskand friends) — use them. This is the preferred path: it needs no shell, and it is the only one that survives a very long generation. If a call fails to authenticate — rather than failing on its inputs — this transport is not usable in this session: go to 2 instead of retrying it. - No usable Sonilo MCP tools, but
sonilo accountexits 0 — use the CLI commands below. Same API, same account, same credential file. Probe withsonilo account, notsonilo whoami: whoami exits 0 even when signed out, so it cannot tell the two states apart. - Neither — stop and run the setup-api-key skill. Do not call
api.sonilo.comwith curl to work around it; both transports handle uploads, polling and retries that a bare request does not.
Quick Start
MCP tool call (recommended)
get_sfx_task(task_id="a1b2c3d4-...")
The tool has two names. The local sonilo-mcp server calls it
get_sfx_task; the hosted OAuth server (the sonilo-claude-plugin path) calls
it get_generation_task(task_id). Same job, same arguments — call whichever one
the connected server actually exposes rather than assuming, since only one of
the two exists in any given session.
Python (pip install sonilo)
from sonilo import Sonilo
client = Sonilo() # reads SONILO_API_KEY
status = client.tasks.get("a1b2c3d4-...") # single check, never raises on failure
result = client.tasks.wait("a1b2c3d4-...") # polls until terminal; raises TaskFailedError/TaskTimeoutError
result.save("recovered.wav")
tasks.get/tasks.wait default to parsing an SFX-shaped result. For a task from dubbing, video_to_sound/video_to_video_sound, analyze_video, or async video_to_music, pass the matching parser explicitly (from sonilo.resources.tasks import parse_music_result, parse_sound_result, parse_dubbing_result, parse_video_analysis_result) — see sonilo-python's README for the exact call shape per task type.
JavaScript / TypeScript (npm install sonilo)
import { SoniloClient } from "sonilo";
const client = new SoniloClient(); // reads SONILO_API_KEY
const status = await client.tasks.get("a1b2c3d4-..."); // single check
const result = await client.tasks.wait("a1b2c3d4-...", { pollInterval: 2000, timeout: 600_000 });
Same caveat as Python: pass the task's actual result type as wait<T>()'s generic (e.g. wait<MusicTaskResult>(...)) for a non-SFX task — see sonilo-js's README.
CLI (npm install -g sonilo-cli or pip install sonilo-cli)
sonilo tasks get a1b2c3d4-...
# npm's sonilo-cli: --poll-interval/--timeout are milliseconds
sonilo tasks wait a1b2c3d4-... --poll-interval 2000 --timeout 120000
# pip's sonilo-cli: the same flags are seconds
sonilo tasks wait a1b2c3d4-... --poll-interval 2 --timeout 600
The two sonilo-cli packages (npm and pip) aren't the same codebase — double-check which one is installed before assuming a flag's units.
cURL (raw REST API, no MCP host)
curl "https://api.sonilo.com/v1/tasks/a1b2c3d4-..." \
-H "Authorization: Bearer $SONILO_API_KEY"
The MCP tool does one status check and, if status is terminal, downloads and saves the resulting file(s) for you — it does not poll in a loop.
Tool
| Tool | Description |
|---|---|
get_sfx_task(task_id, output_directory?) | Check a task; if it succeeded, download and save its result file(s). |
Behavior by Status
| Status | What happens |
|---|---|
processing | Returns a "still processing, try again later" message. No file saved. Call again after a short wait. |
succeeded | Downloads and saves the result — audio for text_to_sfx/video_to_sfx; a single .wav or .mp4 for audio_ducking; a single .mp4 for video_to_video_music/video_to_video_sfx/video_to_video_sound; a single .wav for video_to_sound; one .mp4 per language for dubbing; for a video_to_music(preserve_speech=true) task, the music stream(s) plus the vocals stem plus the mux. An analyze_video task is the exception: it has no file at all, |