Skip to main content
Glama
seocombat

Clipwright

by seocombat
OverviewSchemaSourceRelated ServersDiscussionsAdmin
TypeScript
Remote

Make a talking-actor video

make_ugc

Create a lip-synced UGC ad video by submitting a script and choosing an actor. Starts rendering and returns a run ID to track until the video is ready.

Instructions

Start generation of a lip-synced UGC video. Give a script within the selected speech model's text limit; the actor comes from actor_id (a saved actor from list_actors) or image, otherwise the default actor is used. Format and resolution follow the request and the source, defaulting to 1080x1920. Captions are OPT-IN: ask the user first. Fields the renderer does not honor yet carry a NOT HONORED YET note in their own description — read it instead of guessing. Call quote_ugc before generating and show the cost. This does NOT wait for the video: it starts the run and returns a run_id IMMEDIATELY. You MUST then poll get_run with that run_id until the state is 'succeeded' (video_url) or 'failed'. A 'failed' run whose paid vendor job we still hold can go back to 'queued' and reach 'succeeded' later; whenever that happens it is named in warnings[]. Pass attempt=2,3,… to deliberately start a NEW run for the same input (retry after a failure).

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
lookNoNOT HONORED YET: look is not honored yet: it does not reach the renderernatural
nameNoNOT HONORED YET: name is not honored yet: it does not reach the renderer
imageNoPublic https url of the actor's photo (PNG, JPEG or WebP, up to 10 MB). A file on disk goes through upload_image (POST /v1/uploads) first — pass the url it returns. A source we cannot use — private or loopback host, http, unreachable, redirecting, over 10 MB, or not one of those image types — is refused (unusable_source) before any charge. We do not detect the face's gender: pass actor_gender or voice, or the default male voice is used with a warning.
voiceNoVoice name from list_voices. Curated presets: owner_ru_clone | sarah | george | eric | daria_ru_female (owner_ru_clone is the Russian cloned voice). The API refuses a name list_voices does not return, before any charge. Omitted means the default voice for the actor's gender: the gender of actor_id, actor_gender with image, or george for the default actor and for image without actor_gender. Mutually exclusive with voice_id.
personNoNOT HONORED YET: person is not honored yet: this request uses the default actor; choose actor_id from list_actors or provide image to select a different face
scriptNoThe words the actor says; required unless segments supplies the spoken text. Segments and text-anchored inserts require long-form qualification on the server. Script limits by speech model: eleven_v3: 5000 characters; eleven_flash_v2_5: 10000 characters; eleven_turbo_v2_5: 10000 characters. Count includes spaces, audio tags and stress marks; emoji may count as two characters. There is no word-count limit. Duration and price are estimates until measured. Russian stress: write the stressed vowel as a capital inside a lowercase word ("потОм", "зАмок") and eleven_v3 receives it as the stress mark U+0301 ("пото́м"); a mark typed directly is kept. A capital at the start of a word stays a capital, and a word with a second capital or a capital consonant inside (all caps, "ВУЗы") is left as it is. A single capital vowel inside a word is always read as stress, so write "Яндекс Еда", not "ЯндексЕда". Tell users writing in Russian that they can mark stress this way. eleven_flash_v2_5 and eleven_turbo_v2_5 cost less but misread stress marks: capitals reach them unchanged.
attemptNo
insertsNoText-anchored image inserts over full narration, each covering cover_words spoken words from its anchor; requires long-form qualification on the server, captions=false, 1080p and explicit broll_policy=anyone.
actor_idNoSaved Clipwright actor ID from list_actors. Choose actor_id, image, or person; do not combine them. Without voice or voice_id the voice follows the actor's gender. Do not combine with actor_gender.
captionsNoNOT HONORED YET: captions requested but not rendered in this prototype (stage-B)
segmentsNoOrdered actor and image segments; requires long-form qualification on the server, captions=false and 1080p. Image media requires explicit broll_policy=anyone.
voice_idNoRaw vendor voice id (16–32 letters and digits) for a voice outside the catalog. Checked lazily: an unknown id fails the run, not the request. Mutually exclusive with voice.
tts_modelNoSpeech model: eleven_v3 | eleven_flash_v2_5 | eleven_turbo_v2_5. Omitted means the model of the chosen preset (list_voices shows it; every preset speaks eleven_v3) or eleven_v3 for a raw voice_id. eleven_v3 is the most expressive and the only one that reads stress marks (a capital vowel inside a Russian word, "потОм", becomes one; see script); eleven_flash_v2_5 and eleven_turbo_v2_5 are cheaper alternatives for languages other than Russian. Script limits by speech model: eleven_v3: 5000 characters; eleven_flash_v2_5: 10000 characters; eleven_turbo_v2_5: 10000 characters. Count includes spaces, audio tags and stress marks; emoji may count as two characters. There is no word-count limit. Duration and price are estimates until measured.
backgroundNo
resolutionNoOutput resolution: 720p | 1080p | 4k (short side 720 / 1080 / 2160 px). Omitted means 1080p.
actor_genderNoGender of the face in image: female | male. Only with image: picks the default voice of that gender (female: sarah, male: george). Refused with actor_id (its gender is known) and without image. An explicit voice or voice_id wins and the response warns that actor_gender changed nothing.
aspect_ratioNoOutput format: 9:16 | 1:1 | 16:9. Omitted means 9:16, and a source of another shape is snapped to 9:16 with a warning — pass it explicitly whenever you pass image. A mismatch above 15% between the request and the source is refused (aspect_conflict) before any charge.
broll_policyNoSTORED ONLY: Saved policy for B-roll: anyone allows people including the actor; no_actor excludes the actor; no_people excludes all people, including hands. Segmented media generation is closed. This setting is stored only and has no effect on actor-only videos. Run override wins over the account actor default; otherwise no_people.
caption_styleNoNOT HONORED YET: caption_style is not honored: captions are not rendered in this prototype (stage-B)hormozi
disclosure_overlayNo

Schema Changelog

Changes observed during successful MCP inspections.

  1. First observedv0.20.0

TDQS

A4.9/5.0
Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

The description clearly discloses that the tool is asynchronous: it starts the run and returns a run_id immediately without waiting for the video. It also explains that failed runs may later requeue and succeed, that retries should use attempt=2,3,..., and that certain parameters are not honored yet. This goes far beyond the minimal readOnlyHint/destructiveHint annotations and gives the agent an accurate mental model of the tool's behavior.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Though the description is long, the tool is genuinely complex with 20 parameters and asynchronous behavior. The most important facts are front-loaded: what it creates, how the actor is chosen, the default resolution, and the immediate-return behavior. Every sentence adds operational value, including the warning about NOT HONORED YET fields and the retry semantics.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given the tool's complexity, absent output schema, and sparse annotations, the description is remarkably complete. It covers quoting, polling, retries, warnings, unsupported fields, default behaviors, error conditions that avoid charges, and when to use upload_image. An agent has everything needed to invoke make_ugc correctly and follow up on the run.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

The input schema already covers 85% of parameters in detail, so the baseline is 3. The description adds meaningful cross-parameter guidance: script limits depend on the speech model, actor comes from actor_id/image or defaults, format and resolution default to 1080x1920, captions are opt-in, and attempt starts a new run. This supplements the schema rather than repeating it, which justifies a score above baseline.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description opens with a specific verb and resource: 'Start generation of a lip-synced UGC video.' It clearly names the actor inputs (actor_id, image, or default actor) and distinguishes the tool's role from siblings like quote_ugc and get_run. An agent can immediately understand what make_ugc does and how it differs from related tools.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines5/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description gives explicit operational guidance: call quote_ugc first and show the cost, poll get_run with the returned run_id until succeeded or failed, and use upload_image for files on disk. It also tells agents that captions are opt-in and to ask the user first, and to read NOT HONORED YET field descriptions rather than guessing. This is exceptional when-to-use and how-to-sequence guidance.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Deploy Server

Other Tools