/generate-assets
Agent 6: Generation Unit (any industry)
Agent 6, Generation Unit: resolves slot: ids into real media — B-roll, voiceover, cloned voices, dubs, avatars, mattes, reframes, upscales, 3D stingers, thumbnails, brand assets — one asset per slot with provenance, aigc:true, and credits quoted AND billed. The only skill in the squad that spends credits on media, and never without a live quote against the standing approval. An AI presenter from a script through HeyGen (C51) renders only after the likeness-consent gate passes and lands with aigc: true.
builtv0.2.0
Its status above is the registry's own, with the date of the run that produced it.
What it is for
Contract A
This agent owns no numbered ledger item of its own.
Example requests
- “generate b-roll of”
- “voiceover for this script”
- “clone this voice”
- “dub into”
- “remove the background”
- “a 9:16 without cropping”
- “3d logo sting”
- “upscale to 4k”
- “generate the logo with alpha”
- “an avatar presenter reading this script”
What it needs, and what it hands back
Needs connected
Nothing. It runs on what SHAI already knows about you.
Better with: openart, higgsfield, heygen.
Uses openart_model_form_get, openart_model_cost, openart_generate_video, openart_generate_image, openart_creation_get, generate_video (get_cost:true), generate_video_batch, jobs_wait, show_generation_by_ids, generate_audio, create_voice_from_confirmed_audio, list_voices, voice_change, dubbing, remove_background, reframe, upscale_video, upscale_image, outpaint_image, generate_image, generate_3d, scene_builder_3d_*, apps_describe, apps_invoke, media_upload, media_confirm, media_import_url, balance, transactions, Bash (ffmpeg normalization, dub mux, procedural SFX, zoompan Ken Burns), heygen_video_create, heygen_video_status, heygen_video_fetch, heygen_avatars_read, heygen_voices_read, tools/likeness-consent.py, which exist in Claude Code. In the platform those steps come back as something for you to do, with the exact clicks.
Hands back
- E/generated/<kind>/<slot-id>.<ext> at the requested spec
- E/generated/manifest.json (quoted and billed credits, provenance, qc, aigc:true, voices[])
- E/generated/dub/<lang>.mp4
- brand assets copied into clients/<slug>/assets/brand/ with their rights rows
- rights.md rows and status.md ## Credits (honest accounting) lines
Stops and asks
- the standing credit approval in brief.md before any paid call; above it, stop and come back once with the number
- written consent plus its own approval line before any voice clone (charged in full on creation)
- an unlim_choice return is never answered by this skill — it goes to the user
- no apps_invoke action annotated paid or destructive outside the approval
Steps it cannot do through an API, and hands you instead: SFX Generator and Music Generator are manual-fallback — no generator exists on either provider; the client supplies a licensed file with its rights.md row; remove_background returns an opaque black composite, not an alpha matte — for a true matte, shoot against a key colour and /finish-edit chroma-keys it; TTS, 3D, lip-sync and motion transfer are needs-probe (ceiling reached) — they need a fresh Higgsfield ceiling before use; EGRESS_BLOCKED or no credit -> the slot parks with its prompt intact and the cut holds on the previous shot; HeyGen digital twin of a real person: created in HeyGen's own app with the subject's on-camera consent (twin creation via API is Enterprise-gated), then referenced by avatar_id; the likeness row in rights.md is written before any render.
How it runs
These are the headings of its own written instructions — 319 lines you can read and change.
- Step 0 — Environment check + inputs + project folder
- Step 1 — Quote every slot, then get one approval
- Step 2 — Probe first, then produce
- Step 3 — QC, record, hand off