IMA AI Music Generator — Suno, DouBao

Generate voiceovers, narration, and spoken audio for videos, explainers, ads, and social content.

Security Scan

The skill's requirements, instructions, and included script are consistent with a text-to-music API client that uses a single IMA API key and api.imastudio.com; nothing indicates disproportionate access or hidden endpoints.

Purpose & Capability

Overall coherent: the skill claims to generate music/voice and the code and SKILL.md implement a text_to_music flow (product list → create task → poll). Minor mismatch: some metadata/homepage references use imaclaw.ai while the API host is api.imastudio.com (inconsistent branding), and the description mentions 'voiceovers/narration' while the code fixes TASK_TYPE to text_to_music (models support vocals for songs but this is not a general TTS endpoint). These are plausible explanations (branding/terminology slop) but worth verifying with the publisher.

Instruction Scope

SKILL.md instructs only calls to api.imastudio.com and UX behavior (progress updates, not exposing internal params). The included script builds requests to the declared API, uses the IMA_API_KEY only for Authorization headers, polls task detail, and returns remote audio URLs. It does not instruct reading unrelated files or other environment secrets.

Install Mechanism

No install spec; the skill is instruction+script only and requires python3 and the 'requests' package. No downloads from arbitrary URLs or extraction operations are present.

Credentials

Only a single credential (IMA_API_KEY) is required and used as the Authorization bearer token to api.imastudio.com. The key is the declared primary credential and no other secrets or host credentials are requested. This is proportionate for an API client.

Persistence & Privilege

The skill does not request persistent or system-level privileges (always:false). The manifest and script indicate no local read/write of preferences or logs by default.

Assessment

This skill appears to be what it says: a Python client for IMA's text-to-music API that requires one API key. Before installing, verify the publisher/site (imaclaw.ai vs imastudio.com branding mismatch) and confirm you trust that provider. Limit risk by using a scoped/test API key (not a key granting broad account rights), review the provider's data/credit usage and privacy policies, and test with minimal inputs. If you need true TTS voiceovers (speech synthesis) rather than song vocals, confirm the API supports that use case — the script is fixed to a text_to_music task and may not perform general TTS.

IMA Voice AI — Music Generator

For complete API documentation, security details, all parameters, and Python examples, read SKILL-DETAIL.md.

Model ID Reference (CRITICAL)

Use exact model_id from this table. Do NOT infer from friendly names.

Friendly Name	model_id	Notes
Suno	sonic	✅ Default, full songs with vocals
DouBao BGM	GenBGM	⚠️ Instrumental only, background music
DouBao Song	GenSong	⚠️ Songs with vocals

User input aliases: BGM/背景音乐/纯音乐 → GenBGM · 歌曲/人声/Song → sonic or GenSong · 默认 → sonic

Music Generation Mode

User intent	model_id	When to use
Background music, instrumental	GenBGM	"做一段BGM" / "纯音乐" / "背景音乐"
Song with vocals	sonic	"写首歌" / "带人声" / "歌曲"
Song (alternative)	GenSong	"豆包歌曲" / "GenSong"

Model Selection Priority

1. User preference (if explicitly stated) → highest priority
2. Fallback default: sonic (Suno)

Task	Default Model	model_id	Notes
General music	Suno	sonic	Full songs, vocals
Instrumental/BGM	DouBao BGM	GenBGM	No vocals
Chinese songs	DouBao Song	GenSong	Alternative to Suno

Script Usage

# Generate music (default: sonic/Suno)
python3 {baseDir}/scripts/ima_voice_create.py \
  --model-id sonic \
  --prompt "upbeat lo-fi hip hop, 90 BPM, no vocals" \
  --output-json

# List available models
python3 {baseDir}/scripts/ima_voice_create.py --list-models

# Generate BGM
python3 {baseDir}/scripts/ima_voice_create.py \
  --model-id GenBGM \
  --prompt "calm piano background music for meditation" \
  --output-json

Sending Results to User

# ✅ CORRECT: Use remote URL directly for inline audio display
message(action="send", media=audio_url, caption="✅ 音乐生成成功！\n• 模型：[Name]\n• 耗时：[X]s\n• 积分：[N pts]\n\n🔗 原始链接：[url]")

# ❌ WRONG: Never download to local file

UX Protocol (Brief)

1. Pre-generation: "🎵 开始生成音乐… 模型：[Name]，预计[X~Y]秒，消耗[N]积分"
2. Progress: Every 30-60s: "⏳ 正在生成中… [P]%" (cap at 95%)
3. Success: Send audio via media=audio_url + include link in caption
4. Failure: Natural language error + suggest alternative models. See SKILL-DETAIL.md for error translation.

Never say to users: script names, API endpoints, attribute_id, technical parameter names. Only: model name · time · credits · result · status.

Environment

Base URL: https://api.imastudio.com Headers: Authorization: Bearer $IMA_API_KEY · x-app-source: ima_skills · x_app_language: en

Core Flow

1. GET /open/v1/product/list?app=ima&platform=web&category=text_to_music → get attribute_id, credit, model_version
2. POST /open/v1/tasks/create → get task_id
3. POST /open/v1/tasks/detail → poll every 5s until resource_status==1

MANDATORY: Always query product list first. attribute_id is required.

Defaults and Timeouts

• Task type: text_to_music (fixed)
• Poll interval: 5 seconds
• Max poll wait: 8 minutes
• Default model: sonic (if --model-id omitted)

Estimated Generation Time

Model	Estimated Time	Poll Every
Suno (sonic)	60~180s	5s
DouBao BGM (GenBGM)	30~90s	5s
DouBao Song (GenSong)	60~120s	5s