Unattended Live Avatar

---
name: "unattended-live-avatar"
description: "Turn one shop portrait and short welcome, product, FAQ, and close scripts into talking-avatar clips a store can loop overnight. This unattended live avatar and night-shift avatar studio can clone or pick a voice, then produce a talking avatar livestream set for welcome, product, FAQ, and close so an unattended livestream or digital human livestream can keep a shop loop presenter on camera. Use it for overnight digital human clips, live room avatar loops, and a night-shift talking-head that keeps the room open."
---

# Unattended Live Avatar

Turn one authorized likeness and voice into a night-shift set of
loopable talking-head clips. The first run may create the voice.
Later nights reuse it.

## Scope and adjacent routes

Use this Skill when a shop wants unattended or overnight talking-avatar
clips it can loop: welcome, product, FAQ, and close.

Route weekly founder thought-leadership to `founder-ip-avatar-studio`.
Route a one-off presenter clip with no loop plan to
`talking-avatar-video`. Route scripts, cards, or demo audio without a
talking-head to `live-commerce-script-studio`. Route a lecture series
with many lessons to `course-video-studio`. Without authorization for
the likeness and the voice, stop at confirmation and do not generate.

## Inputs and defaults

The hard inputs are:

- one accessible portrait the host Agent can inspect;
- either an approved speech track or short scripts plus a voice path;
- whether the user owns or has been granted rights to the likeness and the
  voice.

Reuse language, destination, framing, slot labels, and the frozen
`voice_id` from earlier nights. Default to the welcome, product, FAQ,
and close slots the user asked for, `model: "auto"` for video, the
portrait as the strict first frame, and a source-derived aspect ratio.
Ask only for a missing hard input or an unstated slot the user did not
already name.

## Golden path

1. Inspect the portrait. Record MIME type, width, height, aspect ratio, byte
   size, and whether it has an alpha channel. Confirm likeness rights. For a
   local portrait or sample, upload only through the bundled client after
   inspection. Never pass a local path to a remote tool.
2. If the shop still needs a cloned voice, confirm consent first. Access
   to a file is not consent. A suitable sample is currently about 10 to
   300 seconds, no larger than 20 MiB, and clean single-speaker speech.
   Call `beatra.models.list` with `{"capability":"voice_clone"}` and show
   the clone admission card before any clone `client_request_id` or
   `beatra.voices.clone` call: route `voice_clone`, tool
   `beatra.voices.clone`, live price for one successful activated voice,
   the provisional estimate, the 600-credit signup gift usually cannot
   start this clone, and what happens if the balance is short. Do not
   create `client_request_id` or submit until the user confirms they have
   topped up or already have enough credits. Do not offer a free clone.
   Submit `beatra.voices.clone` exactly once:

   ```json
   {
     "sample": {"type": "artifact", "artifact_id": "artifact_..."},
     "display_name": "Shop Night Voice",
     "consent_attested": true,
     "model": "auto",
     "client_request_id": "opaque-unattended-clone"
   }
   ```

   Poll with `beatra.tasks.get` until terminal, then freeze the returned
   `voice_id`. Otherwise call `beatra.voices.list` when a catalog voice
   still needs to be selected. Never put a display name in `voice`.
3. If the user already supplied an approved speech track for a slot, skip
   synthesis for that slot and admit that audio for video. For a script
   route, call `beatra.models.list` with
   `{"capability":"text_to_speech"}` and `{"capability":"image_to_video"}`.
   Confirm that a current video card admits `[image, driving_audio]`. Compare
   the portrait's actual MIME, dimensions, byte size, and alpha-channel
   presence with every advertised image constraint. Confirm planned speech
   can fit a video duration the route can fully contain. Live driving-audio
   i2v (wan2.7-i2v) SKUs are 2–15s. If any script is longer than the longest
   containable clip, split it into sequential containable segments. Each
   segment is its own paid speech call and its own paid video call, each
   with a new `client_request_id`. Do not shorten words to cheapen a clip.
   Use `mp3` only when the live speech card supports it and the live video
   card accepts `audio/mpeg`. If the user requested `flac`, `opus`, or `pcm`
   and the video route does not accept the resulting format, explain the
   incompatibility before any paid call and obtain the user's choice. Do
   not silently change the format. If any required media fact is
   unavailable or incompatible, stop before TTS.
4. On the script route only, show the narration arguments and paid
   boundary. Freeze the script, voice, format, and one speech
   `client_request_id` per segment. Submit `beatra.speech.synthesize`
   exactly once per segment through the bundled client. Skip this step
   when approved speech already exists for that segment.
5. For each speech segment, or for the supplied approved track, poll
   until terminal audio facts exist. On the script route, poll each
   narration task with `beatra.tasks.get`. Read the returned artifact
   plus actual `task.output.audio.mime_type`, `duration_seconds`, and
   `size_bytes` when present. Approve the real audio before that
   segment's video. Never treat a script preview, expected duration, or
   task metadata as an audio review.
6. For each approved audio segment, refresh the `image_to_video` cards
   and admit the actual portrait plus that audio. Recheck every image
   fact and compare the audio's actual MIME, duration, and byte size
   with the current driving-audio constraints. If terminal audio size
   is absent, obtain it from trusted artifact metadata; if it remains
   unavailable, stop before video. Use the smallest supported integer
   video duration at or above the actual speech length. Show the video
   admission card before any video `client_request_id` or
   `beatra.videos.animate` call: route `image_to_video`, tool
   `beatra.videos.animate`, approved portrait and speech artifacts,
   prompt, audio-led duration, resolution if set, output count,
   provisional live estimate, the fact that the 600-credit signup gift
   usually cannot start this video, and what happens if the balance is
   short. Do not submit until the user confirms they have topped up or
   already have enough credits. Approved narration does not authorize
   the video call.
7. Submit `beatra.videos.animate` exactly once per approved audio
   segment. Do not configure a host Beatra Connector. Do not use
   REST/OpenAPI as a fallback. Poll each video task with
   `beatra.tasks.get` until terminal.
8. Deliver the returned clips in slot and script order with actual
   dimensions, duration, usage, and `billing.net_charged_credits`.
   Review identity, clarity, lip timing, and must-keep drift. Label
   welcome, product, FAQ, and close so the shop can loop them.

## Decisions that require confirmation

Likeness and voice rights, clone admission, narration, and the video
admission card each require their own confirmation.

## Recovery

Each paid stage has its own frozen payload and ID. Recover a lost create
response only with that stage's identical payload. Recover a lost task
ID through `beatra.tasks.list` and `beatra.tasks.get`. Call
`beatra.tasks.cancel` only when the user asks to cancel that stage; on
409 keep polling. On `insufficient_balance`, keep the top-up URL inside
the balance error exact and retry the same frozen ID only after the user
says they have topped up. If a create returns no `task_id`, do not poll;
reconcile before minting a new ID.

## Account balance

When the user asks how many credits remain or whether a live estimate fits,
call `beatra.wallet.get`. When they ask what was charged, call
`beatra.wallet.ledger`. Both are read-only. Do not invent an account-balance or
top-up tool. Do not make `wallet.get` a required step before every paid submit.

When a model card comes back carrying a `top_up` block, relay its tiers as the
card lists them and in that order. Do not rank them, do not talk one down, and
do not pick one for the user. Which tier suits them is their call, made on
the wallet page with the whole list in front of them. Never quote a tier from
memory.

## References by task

- Read [Unattended live avatar workflow](references/workflow.md) for clone,
  narration, admission cards, payloads, and review.
- Read [Installation and authentication](references/installation-and-auth.md)
  only when authorization needs attention.
- Read [Installation registration](references/installation-registration.md)
  for first-use registration.
- Read [Tasks and results](references/tasks-and-results.md) and
  [Billing, errors, and recovery](references/billing-errors-and-recovery.md)
  for task and billing facts.
- Read [Bundled MCP Client diagnostics](references/mcp-connection.md) when
  the bundled client cannot connect.
- Read [Automatic updates and safety](references/automatic-updates-and-safety.md)
  for update controls.
- Read [Uninstall and disconnect](references/uninstall-and-disconnect.md)
  only when the user asks to remove the package.

## Runtime and safe automatic updates

Use or invoke the bundled `scripts/mcp_client.py` for every Beatra operation.
Before ordinary commands it silently checks for a newer release at most once
every 24 hours per installation. Silent checks are enabled by default, and a
newer release installs without separate confirmation.

The updater accepts only the fixed official discovery address and immutable
Beatra CDN path embedded for this package, channel, and locale. It verifies
the discovery data, archive, manifest, and every file's size and checksum
before replacement. It replaces only package-owned files and rejects
redirects, downgrades, wrong package/channel/locale/version data, unexpected
URLs, unsafe archives, and files outside the owned destination.

Update checks, downloads, verification, replacement, rollback, and recovery
fail open: the current installation remains usable and the user's original
command continues. An update failure never authorizes retrying a paid
generation. The automatic-update choice persists across later commands for
this installation:

```text
python3 scripts/mcp_client.py update --auto off
python3 scripts/mcp_client.py update --auto on
python3 scripts/mcp_client.py update --check
```