stepace-experimental（阶梯实验）

StepAce Experimental — AI Music Generation

Queue or schedule AI-generated songs directly to your Android phone through the StepAce Experimental app.

🚀 Setup (first time only)

Before using this skill, you need to connect it to your StepAce Experimental app:

1. Open the StepAce Experimental app on your Android phone
2. Go to Settings (bottom tab)
3. Tap OpenClaw Bridge
4. Tap Connect — the app generates a unique pairing token
5. Copy the pairing token shown on screen
6. Tell your agent:

> *"Set my StepAce token to <paste token here>"*

Your agent saves it as STEPACE_TOKEN. You only need to do this once — unless you regenerate your token in the app.

Token management

• *"Set my StepAce token to XYZ"* → save as STEPACE_TOKEN
• *"Reset my token"* → remind user: StepAce Experimental app → Settings → OpenClaw Bridge → Connect → paste new token
• *"What's my token?"* → confirm it's set, show only first 4 + last 4 chars (e.g. T2L_****IP0)
• If STEPACE_TOKEN is not set when the user asks to generate music → walk them through setup before proceeding

Generation types

Immediate — enqueue_generation

Queues a song to generate right now on the phone. Use this by default.

Scheduled — schedule_generation

Queues a song to generate at a specific future time. Use this when the user gives a time or date ("tonight at 2am", "schedule for tomorrow morning", etc.). Requires scheduledAt as a Unix timestamp in milliseconds.

Parameters exposed to the user

Only caption is required. All others are optional — omit them entirely from the payload if not specified (do not send null or empty values).

Instrumental logic

• User provides lyrics → instrumental: false, include lyrics
• User says "no vocals" / "instrumental" / "beat" → instrumental: true, omit lyrics
• User provides neither → omit both instrumental and lyrics (let the app decide)

API call

Endpoint: POST https://openclaw-bridge.torrico-villanueva-cesar-kadir.workers.dev/openclaw/queue Header: Content-Type: application/json

Preferred transport

Use curl from the shell as the default/preferred way to call the bridge. Do not prefer Python urllib/generic HTTP clients when curl is available, because the bridge/CDN may treat those clients differently and reject them even when the same payload works via curl.

Immediate generation

Preferred example using curl:

source /home/deploy/.stepace-env

curl -X POST \
  'https://openclaw-bridge.torrico-villanueva-cesar-kadir.workers.dev/openclaw/queue' \
  -H 'Content-Type: application/json' \
  --data '{
    "pairingToken": "'"$STEPACE_TOKEN"'",
    "type": "enqueue_generation",
    "payload": {
      "requestJson": {
        "caption": "cinematic synthwave with huge drums",
        "instrumental": true,
        "bpm": 120,
        "duration": 30
      }
    }
  }'

Equivalent JSON payload:

{
  "pairingToken": "{STEPACE_TOKEN}",
  "type": "enqueue_generation",
  "payload": {
    "requestJson": {
      "caption": "cinematic synthwave with huge drums",
      "instrumental": true,
      "bpm": 120,
      "duration": 30
    }
  }
}

Scheduled generation

Preferred example using curl:

source /home/deploy/.stepace-env

curl -X POST \
  'https://openclaw-bridge.torrico-villanueva-cesar-kadir.workers.dev/openclaw/queue' \
  -H 'Content-Type: application/json' \
  --data '{
    "pairingToken": "'"$STEPACE_TOKEN"'",
    "type": "schedule_generation",
    "payload": {
      "requestJson": {
        "caption": "dark techno with metallic percussion",
        "instrumental": true
      },
      "scheduledAt": 1775120400000
    }
  }'

Equivalent JSON payload:

{
  "pairingToken": "{STEPACE_TOKEN}",
  "type": "schedule_generation",
  "payload": {
    "requestJson": {
      "caption": "dark techno with metallic percussion",
      "instrumental": true
    },
    "scheduledAt": 1775120400000
  }
}

scheduledAt must be a Unix timestamp in milliseconds (13 digits). Convert from the user's stated time using their local timezone if known, otherwise ask.

With lyrics

{
  "pairingToken": "{STEPACE_TOKEN}",
  "type": "enqueue_generation",
  "payload": {
    "requestJson": {
      "caption": "fast energetic electronic anthem with punchy drums",
      "lyrics": "We light the night, we never slow, hearts on fire, we steal the show.",
      "vocal_language": "en",
      "instrumental": false,
      "duration": 30,
      "bpm": 160
    }
  }
}

Success response

{
  "jobRef": "ref_abc123...",
  "status": "queued",
  "type": "enqueue_generation"
}

Reply to the user with a message like:

🎵 Song queued on StepAce Experimental!
Caption: <caption>
📅 Scheduled for: <human-readable time>   ← only if scheduled
Settings: 160 BPM · C minor · 4/4 · 30s  ← only mention fields that were set
Job ref: ref_abc123...

Your phone will notify you when it's done 🎶

Error handling

Natural language examples

• *"Make me a dark synth-wave track at 130 BPM"* → enqueue_generation, instrumental: true, bpm: 130
• *"Generate a cumbia song with these lyrics: [...]"* → enqueue_generation, instrumental: false, vocal_language: "es"
• *"Queue a lo-fi beat, 85 BPM, 60 seconds"* → enqueue_generation, instrumental: true, bpm: 85, duration: 60
• *"Schedule a techno track for 2am tonight"* → schedule_generation, resolve scheduledAt from current time + user's timezone
• *"Make a waltz in A minor"* → timesignature: "3/4", keyscale: "A minor"
• *"Something chill in Spanish"* → vocal_language: "es", infer a relaxed caption
• *"Song about the ocean, no lyrics, key of D minor, 45 seconds"* → instrumental: true, keyscale: "D minor", duration: 45