Skip to main content
Glama

Generate a video from a creative brief

autoshort_generate_chat_video
Idempotent

Generate a narrated short video from a natural-language creative brief, optionally placing up to six user-supplied images as the first scenes; then poll the asynchronous render until complete.

Instructions

Create a video from a natural-language creative brief; AutoShort writes the script from it. To keep an agreed script word for word with your own shots, use autoshort_draft_video instead. To use the user's own pictures ("make a video with these images"), pass image_urls (public https links, up to 6): image 1 becomes scene 1, image 2 scene 2, …; the rest are generated. Naming an image inside the brief does NOT attach it: only image_urls do. If the brief points at a picture you did not pass, the answer carries warnings["IMAGE_IGNORED"] with a warningText (read it to the user), or CHAT_NO_IMAGE_INPUT (retry with image_urls, or proceedWithoutImage: true). Optional captionSecondaryLanguage adds a translated second caption line under the narration captions (e.g. "vi" for English narration with Vietnamese under it); "none" turns it off for this video even if the account default has one; omitted = the account default (off unless the user set one). It costs nothing extra. Costs 40 credits with the default voice (+25 if ElevenLabs narrates); images do not change the price. Rendering is asynchronous: poll autoshort_get_video every 20-30 seconds until status is COMPLETED or FAILED. If get_video returns blockedReason, stop polling and act on it (e.g. top up credits). ROUTING: if the user pasted a cinematic / video-model prompt (camera, lens, lighting, a scene description), do NOT use this tool: it writes its own script and makes narrated still-image scenes. Use autoshort_generate_motion_clip, which sends the prompt as written. Use this tool for narrated stories and facts. videoProvider VEO_3_1_OPERATOR is operator-fulfilled and not guaranteed: it does not turn a prompt into a moving clip.

Tool call template

Replace each placeholder with a value matching the input schema. Only required parameters are included; add optional parameters as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "message": "<message>"
    },
    "name": "autoshort_generate_chat_video"
  }
}

Input Schema

NameRequiredDescriptionDefault
messageYesCreative brief describing the desired video.
artStyleNoArt style returned by autoshort_list_niches.
languageNoNarration language. Omit to detect it from the brief.
image_urlsNoOptional. The user's OWN images to use as the first scenes: image 1 = scene 1, image 2 = scene 2, and so on; the remaining scenes are generated as usual. https URLs only, up to 6 per call, JPEG/PNG/WebP, each at most 10 MB and 25 megapixels; at most 30 images per hour per account. Animated WebP uses its first frame; GIF and SVG are rejected. AutoShort downloads each image once, strips EXIF/GPS, re-encodes it and serves it from its own domain (your URL is never linked into the video). Fast (FFMPEG) multi-scene videos only. The answer states how many images were used and for which scenes; more images than scenes is refused, never silently dropped. Only send images the user owns or may use; each is shown as a still with a slow zoom and is never animated or used to generate a person.
aspectRatioNo
videoProviderNoFFMPEG for fast multi-scene video; VEO_3_1_OPERATOR for cinematic quality when eligible.
idempotencyKeyNoRetry-safe: a repeat with the same key returns the first answer (replayed: true) and creates nothing.
targetDurationNoLength in seconds. With image_urls and no targetDuration, 30 is used (6 scenes) so every image has a scene.
proceedWithoutImageNoSet true to create the video from the text even though the brief points at an image you did not pass in image_urls. Without it, a brief that clearly hands over a file ("the photo I attached", an image URL) is refused with CHAT_NO_IMAGE_INPUT before anything is charged; a softer mention creates the video and returns warnings: ["IMAGE_IGNORED"] + warningText.
captionSecondaryLanguageNoSecond caption line language (translated, shown under the narration captions). "none" = no second line for this video. Omit to use the account default. Same language as the narration = ignored.

Schema Changelog

Changes observed during successful MCP inspections.

  1. First observedv2.3.0

TDQS

A4.8/5.0
Behavior5/5

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

The annotations declare only that this is a non-read-only, non-destructive, idempotent, open-world operation; the description adds the operational facts that matter: 40 credits for the default voice (+25 with ElevenLabs), no extra charge for images, asynchronous rendering at 20-30s poll intervals until COMPLETED or FAILED, stopping on blockedReason, image ingestion rules (EXIF stripping, re-hosting, 30/hour), and the warning codes IMAGE_IGNORED / CHAT_NO_IMAGE_INPUT with their refusal-versus-warn semantics. It also warns that VEO_3_1_OPERATOR is operator-fulfilled and not guaranteed. That warning prevents a mode-selection mistake that the annotations cannot.

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

Conciseness4/5

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

It is front-loaded: purpose, then alternatives, then images, then captions, then cost and async, then routing. Every paragraph earns its place, and the capitalized error codes are easy to scan. It runs long, and the image paragraph packs a lot of detail, such as re-encoding and megapixel limits, into the tool description rather than leaving that to the schema.

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?

There is no output schema, so the description must carry the lifecycle itself, and it does: it tells the agent to poll autoshort_get_video, which terminal statuses to look for, how to react to blockedReason, and what the warnings array means plus how to relay warningText to the user. For a 10-parameter, credit-charging, asynchronous tool, nothing essential to correct invocation is missing.

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?

Schema coverage is already 90%, so the baseline is 3, and the description still adds meaning: it explains that naming an image inside the brief does NOT attach it and only image_urls does; that image 1 becomes scene 1 with the rest generated; that 'none' for captionSecondaryLanguage disables it even when the account default has one, while an omitted value means the account default; and that VEO_3_1_OPERATOR will not turn a prompt into a moving clip. The remaining parameters, such as artStyle, aspectRatio, and targetDuration, carry their meaning in the schema alone.

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?

States a specific verb and resource: 'Create a video from a natural-language creative brief; AutoShort writes the script from it.' The clause about the script clarifies the division of labor and separates it from both autoshort_draft_video (which keeps an agreed script word for word) and autoshort_generate_motion_clip (which processes a pasted cinematic prompt). An agent can choose correctly without opening a schema.

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?

It gives explicit alternatives and the conditions that select them: draft_video for keeping an agreed script word for word, motion_clip when the user pasted a cinematic prompt or a video-model prompt, and get_video for polling. The final ROUTING paragraph even states which input is being handled: narrated stories and facts. It also states when the user's own pictures should be routed through image_urls. Neither the 'when' nor the 'when not' is left to inference.

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