beatra-ai/beatra-skills / skills/ai-video-image-music-studio
ai-video-image-music-studio
VerifiedGenerate AI video, images, music, and voice-over in one connected creative flow, edit visual results, and keep everything you make easy to find. This all-in-one AI media generator turns text into images, images and references into video, ideas into songs or instrumentals, and scripts into narration, and can build a custom voice when a series needs its own sound. Working as an AI video generator, AI image generator, AI music generator, and AI voice generator in one place, it covers text-to-image, text-to-video, image-to-video, AI video editing, text-to-speech, multilingual voice-over, and voice cloning for social posts, product visuals, ads, courses, podcasts, short films, and cross-media campaigns. Reuse source media and finished files across formats, follow production progress, and keep every finished piece in one place.
references/models.md
View on GitHub# Models
Prefer Beatra's automatic routing. Omit `model`, or leave it as `auto`, unless
the user explicitly names a model or asks to choose, compare, or change models.
When model choice matters, call `beatra.models.list` for the exact generation
capability. Present only the returned selectable models and use its current
constraints, defaults, supported media combinations, and customer price inputs.
Respect an explicit selection and allow validation to reject an incompatible
combination rather than silently substituting another model.
Every returned control name is an exact MCP request path. A dotted path names a
nested object and `[]` names each item in an array, for example
`references[].reference_voice`. Do not flatten or rename these paths.
For image capabilities, the public input and output limit is four. A concrete
model can expose a lower input maximum in its interface card; explicit requests
must respect that value, while `auto` evaluates the complete request and may
select another eligible model. Never truncate, merge, or drop an image. Image
pricing uses `pricing.options`. Select an option only when all of its
`dimensions` match the admitted request; an empty dimensions object is the
default. A preset canvas exposes its `tier` as the `resolution` dimension. When
target-canvas normalization or request-dependent `auto` leaves multiple
eligible prices, quote the returned range and approve against the maximum
rather than guessing. Apply the card's `estimate_formula`; under
`billing_basis: "successful_image_count"`, source-image count never multiplies
the customer estimate and the task's persisted artifacts and terminal billing
are final truth. Never copy model names, tier names, thresholds, or credit
amounts into the Skill as durable catalog data.
For video capabilities, `auto` selection is request-dependent. Discovery's
`auto.candidate_order` is the live priority order; Beatra evaluates the complete
request against that order and selects the first eligible model. The selected
concrete model is persisted on the Task as `resolved_model`. Do not describe
`auto` as one fixed model. If controls are omitted, each candidate is evaluated
with its own returned defaults.
When the user supplies a concrete model, pass that concrete model unchanged.
Do not continue down `candidate_order` and do not silently replace it when the
request is incompatible or temporarily unavailable.
For text-to-speech (`text_to_speech`) and multi-speaker dialogue
(`text_to_dialogue`), customer pricing is linear per 1,000 weighted characters:
each Han ideograph counts as two and every other character counts as one.
Requests below 1,000 are not rounded to a full block. Smart Voice Director
(`speech_script`) bills per call (48-hour identical `input`/`instructions`/`language`/`context`
re-calls swapping only `voice` or `model` add zero charge), sound effects
(`sound_effects`) bill per generated `0.5–30 s` clip, and audio finishing or
music editing bills under its exact capability (`audio_process`,
`audio_stem_separation`, `audio_voice_isolation`, `music_edit`). The
customer-facing task contract remains asynchronous while Beatra performs
automatic execution routing internally. The exact TTS model card's
`constraints.supported_languages` is the source of truth for the
public BCP-47 `language` field; see [speech and voices](speech-and-voices.md) and
[speech and audio](speech-and-audio.md) for current rules. Voice cloning is
priced once per successful activated voice, including activation. Read all
live rates from `beatra.models.list`; do not reuse remembered model or price data.
For a preflight estimate, use the returned price options for the relevant
candidate or concrete model and state that request-dependent auto selection is
finalized at admission. Image cards use their own `estimate_formula` and
dimension-matched `unit_credits`. Other media may use
`(unit_price_credits * billable_quantity) / scale`. For video, read each price
option's `billable_quantity.basis` and `counted_inputs`; when the basis is
`capped_reference_video_seconds`, apply the returned aggregate maximum and
`per_video_cap_seconds_by_video_count` map. For other media, read the returned
`billing_basis`. This is an estimate until actual task usage is known. The
request-time prepaid debit is therefore an estimate for workflows whose
billable duration depends on accepted media or generated output. On terminal
success, use the task's returned usage and billing as truth:
`charged_credits` includes all debits, `refunded_credits` includes any
unused-estimate refund, and `net_charged_credits` is the final charge.
Do not list model catalogs from memory and do not call model discovery as a
routine prerequisite for generation.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack." | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |