heygen

HeyGen Video Agent — NanoClaw Container Skill

When to Use

Use this skill when the user wants to create a video with an AI avatar presenter. Triggers: "make a video", "create a video message", "record a video", "avatar video", "talking head video", "video pitch", "video update".

NOT for: image generation, audio-only TTS, video translation, or cinematic b-roll.

Required Environment

• HEYGEN_API_KEY — Get from https://app.heygen.com/settings?nav=API
• heygen CLI — install: curl -fsSL https://static.heygen.ai/cli/install.sh | bash. Verify: heygen auth status.

Steps

Step 1: Discover Available Avatars

heygen avatar list --ownership public --limit 5 | jq '.data[] | {group_id: .id, avatar_name: .name}'

avatar list returns avatar groups — each .id is a group_id, not an avatar_id. The avatar_id you pass to generation is a specific look: list looks with heygen avatar looks list --group-id <group_id> | jq '.data[] | {avatar_id: .id, preview_image_url}' and pick a look's .id. If the user already has a specific look id, use it directly.

Step 2: Find a Voice

heygen voice list --limit

Pick a voice_id matching the desired language and tone.

Step 3: Write the Script

Write a spoken-word script for the avatar. Rules:

• Write for speech, not text. Short sentences. Natural pauses.
• 150 words per minute is the target pace.
• 30 seconds = ~75 words. 60 seconds = ~150 words.
• No stage directions. No markdown. Just what the avatar says.

Step 4: Generate the Video

With --wait, the CLI blocks until the video completes and emits the final status object. Without --wait, submission returns immediately — save the session_id from stdout for manual polling.

--timeout 10m is sized for short videos (≤60s). Bump to --timeout 45m for videos >60s or when targeting 2+ minute output — Video Agent generation routinely takes 20-45m for longer clips.

Step 5: Poll for Completion (only without --wait)

Poll every 15 seconds. Session status is one of thinking, waiting_for_input, reviewing, generating, completed, failed — not a strict linear sequence (waiting_for_input only occurs in chat mode). Terminal states are completed and failed.

Once .data.video_id is present, run heygen video get <video_id> and read .data.video_url for the download URL (and .data.failure_message on failure).

Step 6: Deliver

Download the video and present it to the user:

Writes the MP4 to disk and emits {"asset", "message", "path"} on stdout — chain on .path.

Verification

After generating a video, confirm:

1. CLI exits 0 (generation accepted). Without --wait, stdout includes a session_id for polling; with --wait, the CLI polls the video to completion and stdout is the final video resource (.data.id, .data.video_url).
2. Polling (or --wait) returns status: "completed" within 5 minutes
3. heygen video get <video_id> returns a valid HTTPS .data.video_url
4. Downloaded file is a playable MP4

Report Feedback

After the video is delivered and verified, report how the run went (once per task):

--rating runs 1 (broke / unusable) to 5 (worked great); put any bug or friction in --comment. Anonymous, no API key; no-ops if analytics are opted out (HEYGEN_NO_ANALYTICS or heygen config set analytics false).

Troubleshooting

Limits

• Free tier: 1 minute of video per month
• API trial: 3 free credits on signup
• Max video length per request: ~5 minutes
• Concurrent generation: depends on plan tier