Intern Onboarding Talking Clips

Turn already-written intern onboarding notes and authorized stills into one talking clip per still. Deliver 2 to 8 clips. Keep each clip as its own file.

Scope and adjacent routes

Use this Skill when a recruiter or hiring manager wants short talking clips that walk an intern through first-day facts from stills they can authorize.

Route a hiring video for an open role to hiring-avatar-studio. Route a generic presenter that is not intern onboarding to talking-avatar-video. Route a homestay welcome to airbnb-welcome-avatar. Route a multi-lesson lecture series to course-video-studio. Route a creator drop teaser to creator-drop-talking.

Collect the onboarding pack

Hard inputs are:

• at least one accessible still the host Agent can inspect — a recruiter or founder portrait, or a workplace photo that will be the first frame;
• confirmed intern onboarding facts the team already wrote (hours, first-day checklist, workplace location, safety notes, next step);
• likeness and voice rights when a face or a cloned voice will appear;
• how many clips the pack should contain, or permission to use the default of 3.

Reuse already-known language, workplace names, and a frozen voice_id. Ask only for a missing hard input. A count outside 2 to 8 is still doable: confirm that pack size and its live cost.

Do not invent a stipend, schedule, badge code, or face. File access is not consent.

Inspect every still. Record MIME type, width, height, aspect ratio, byte size, and whether it has an alpha channel. For a local file, upload only through the bundled client after inspection (scripts/mcp_client.py / beatra.assets.upload). Keep the returned artifact id. Never pass a local path to beatra.voices.clone, beatra.speech.synthesize, or beatra.videos.animate.

Plan the free clip list

Write a labeled intern talking list before any paid clone, speech, or video. Default three clips unless the team names another count in 2 to 8: welcome, first day, and next step. Each item records the still, the spoken line from confirmed handbook facts, intended length as a 2–15s clip, and whether it uses a catalog voice, an approved track, or a clone.

That list is the free visible result. Planning is not approval.

Safe defaults:

• one beatra.videos.animate call per still;
• model: "auto" unless the team chose a live SKU;
• the still as the strict first frame;
• source-derived aspect ratio;
• audio-led duration inside 2–15s. Keep each clip as its own file.

Confirm clone, speech, then video

Clone, speech, and video are separate paid stages. Each stage gets its own six-field card and its own opaque client_request_id per clip.

If the team wants a cloned voice, inspect an authorized sample, read the live voice_clone card, and wait on the clone card before beatra.voices.clone. A found file is not clone consent. Show the clone card and wait:

1. Work — one authorized voice sample (beatra.voices.clone).
2. Credits — the live voice_clone price just read. Do not reuse a remembered number.
3. Count — one paid clone call for this sample.
4. Identity — one new opaque client_request_id.
5. If we stop here — the labeled clip list remains usable.
6. If the balance is insufficient — relay the official message and its top-up URL exactly. Translate the prose; keep the URL. Do not retry until the team says they have topped up.

Before speech, call beatra.models.list for text_to_speech:

{"capability": "text_to_speech"}

Use beatra.voices.list when a catalog voice still needs a choice. Show the speech card and wait:

1. Work — one spoken intern line per named clip (beatra.speech.synthesize).
2. Credits — the live text_to_speech price just read, times the clip count. Do not reuse a remembered number.
3. Count — one paid speech call per clip.
4. Identity — one new opaque client_request_id per clip.
5. If we stop here — the labeled clip list remains usable.
6. If the balance is insufficient — relay the official message and its top-up URL exactly. Translate the prose; keep the URL. Do not retry until the team says they have topped up.

Submit each speech clip once through bundled scripts/mcp_client.py. Send the spoken line in input. Poll beatra.tasks.get. Read actual audio MIME, duration, and size. A script preview is not the audio review.

Before video, call beatra.models.list for image_to_video:

{"capability": "image_to_video"}

Confirm a current card admits [image, driving_audio]. Compare every image fact and the real audio MIME, duration, and size with advertised limits. Live driving-audio clips are 2–15s. If speech is longer than the longest containable clip, split into sequential containable segments; each segment is new speech and new video, each with a new ID. Deliver the segments as separate files. Approved narration does not authorize video. The 600-credit signup gift usually cannot start this video. Show the video card and wait:

1. Work — one intern talking clip per named still (beatra.videos.animate).
2. Credits — the live image_to_video price just read, times the segment count. Do not reuse a remembered number.
3. Count — one paid video call per approved audio segment.
4. Identity — one new opaque client_request_id per segment.
5. If we stop here — the clip list and approved narration remain usable.
6. If the balance is insufficient — relay the official message and its top-up URL exactly. Translate the prose; keep the URL. Do not retry until the team says they have topped up.

Then submit beatra.videos.animate once per approved segment.

Review, deliver, and recover

Review identity, speech clarity, and mouth timing. Report only what the host can actually see and hear. Do not promise perfect lip sync. Never invent a stitch, concat, or editor tool. After each terminal paid task, deliver actual bytes plus MIME, duration, and size when present, and billing.net_charged_credits. Do not promise the prepaid estimate is the final charge.

After a returned task_id, poll that task. If the create response is lost, search with beatra.tasks.list and verify with beatra.tasks.get before replay. Reuse an ID only with byte-identical arguments. A changed still, line, voice, or duration is a new card and a new ID. Cancel only when the team asks.

Execution

Invoke every remote Beatra operation only through this package's bundled scripts/mcp_client.py. Put the MCP tool name after call and send one JSON object on standard input.

python3 scripts/mcp_client.py call beatra.models.list

{"capability": "text_to_speech"}

python3 scripts/mcp_client.py call beatra.models.list

{"capability": "image_to_video"}

printf '%s' '{"image":{"type":"artifact","artifact_id":"art-intern-01"},"driving_audio":{"type":"artifact","artifact_id":"art-speech-01"},"prompt":"A restrained intern welcome with steady eye line and a stable camera.","duration":8,"client_request_id":"opaque-intern-video-01"}' | python3 scripts/mcp_client.py call beatra.videos.animate

Do not configure or call a host Beatra Connector, and do not use REST/OpenAPI as a fallback.

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

• For clip lists, payloads, and recovery, read Intern onboarding talking workflow.
• For authorization and the non-billable registration step, read installation and authentication and installation registration.
• For shared task, billing, and connection details, read tasks and results, billing, errors, and recovery, and Bundled MCP Client diagnostics.
• For update guarantees and controls, read automatic updates and safety. For removal, read uninstall and disconnect.

Runtime and safe automatic updates

The bundled client silently checks for a newer release at most once every 24 hours per installation. When a newer version is available, it installs automatically without separate confirmation. It downloads only from the fixed official Beatra discovery and immutable CDN paths for this package, channel, and locale, verifies discovery data, archive, manifest, and every packaged file, and replaces only package-owned files.

Update checks, downloads, verification, replacement, rollback, and recovery fail open: the current installation remains usable and the original command continues. An update failure never authorizes retrying a paid clone, speech, or video request. The setting persists for this installation. See automatic updates and safety.

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