简体中文 | English

Giggle Music

Source: giggle-official/skills · API: giggle.pro

Generates AI music via giggle.pro. Supports simplified and custom modes. Submit task → query when ready. No polling or Cron.

API Key: Set system environment variable GIGGLE_API_KEY. Log in to Giggle.pro and obtain the API Key from account settings.

Important: Never pass GIGGLE_API_KEY in exec's env parameter. API Key is read from system environment variable.

No Retry on Error: If script execution encounters an error, do not retry. Report the error to the user directly and stop.

Interaction Guide

Mode Selection (priority: high to low)

Key rule: If the user does not provide lyrics, always use simplified mode A. Use the user's description exactly as --prompt; do not add or rewrite. E.g. user says "female voice, 1 min, ancient romance", use --prompt "female voice, 1 min, ancient romance" directly.

Guidance when info is lacking

Only when the user input is very vague (e.g. "generate music" with no description), ask:

Question: "What type of music would you like to generate?"
Options: AI compose (describe style) / Use my lyrics / Instrumental

Execution Flow: Submit and Query

Music generation is asynchronous (typically 1–3 minutes). Submit a task to get task_id, then query when the user wants to check status.

Step 1: Submit Task

First send a message to the user: "Music generation submitted. Usually takes 1–3 minutes. You can ask me about the progress anytime."

A: Simplified Mode

python3 scripts/giggle_music_api.py --prompt "user description"

B: Custom Mode

python3 scripts/giggle_music_api.py --custom \
  --prompt "lyrics content" \
  --style "pop, ballad" \
  --title "Song Title" \
  --vocal-gender female

C: Instrumental

python3 scripts/giggle_music_api.py --prompt "user description" --instrumental

Response example:

{"status": "started", "task_id": "xxx"}

Store task_id in memory (addMemory):

giggle-generation-music task_id: xxx (submitted: YYYY-MM-DD HH:mm)

Step 2: Query When User Asks

When the user asks about music progress (e.g. "is my music ready?", "progress?"), run:

python3 scripts/giggle_music_api.py --query --task-id \x3Ctask_id>

Output handling:

Link return rule: Audio links in stdout must be full signed URLs (with Policy, Key-Pair-Id, Signature query params). Correct: https://assets.giggle.pro/...?Policy=...&Key-Pair-Id=...&Signature=.... Keep as-is when forwarding.

Recovery

When the user asks about previous music progress:

1. task_id in memory → Run --query --task-id xxx directly. Do not resubmit
2. No task_id in memory → Tell the user, ask if they want to regenerate

Parameter Reference