Tools

Every tool of the Argil MCP server, its arguments and what it returns.

Your assistant reads these tools, and their exact arguments, from the server: you never call them yourself. This page is for checking what an assistant can do, and for building agents.

Tool	Does	Main arguments
search_models	Models your workspace can use, with modes and prices	query, type (image, video, audio), mode
get_model	One model; with a mode, its exact fields, limits and price	model, mode
estimate_generation	The price of a request, and whether the balance covers it	model, mode, input
generate_media	Start a generation. Spends credits	model, mode, input, max_credits, idempotency_key
get_generation	Status, cost and outputs; can wait	id, wait_seconds
list_generations	Past generations, newest first	model, mode, created_after, metadata
delete_generation	Remove a finished generation (outputs stay in the library)	id
add_asset	Add a file to the library	url, file or upload
get_asset	One file; can wait until it is ready	id, wait_seconds
list_assets	Files of the library: uploads, imports and outputs	type
delete_asset	Remove a file	id
list_elements	Characters, styles, locations, props and lore, with their plates	kind, ids
list_voices	Voices to speak with: yours and Argil’s	visibility, language, gender
list_avatars	Avatars to animate, with their image: yours and Argil’s	visibility, orientation
get_balance	Credits available and reserved
list_webhook_endpoints, create_webhook_endpoint, update_webhook_endpoint, delete_webhook_endpoint	URLs Argil calls, with a signed POST, when a generation succeeds or fails	url, enabled_events

Lists take limit (1 to 50) and starting_after to page. Only generate_media spends credits. The delete_* tools, and create_webhook_endpoint and update_webhook_endpoint (which send your outputs’ links to a URL), are marked destructive, so clients ask before running them. add_asset, estimate_generation and generate_media, which can download a URL you pass, and the two webhook tools are marked open-world. Names and descriptions in results are shown as quoted strings: text your team wrote, which the assistant treats as data. Creating a webhook endpoint, giving one a new URL or turning one back on also needs your own confirmation in a dialog your client shows (MCP elicitation, protocol 2026-07-28, as in Claude Code and claude.ai): the assistant cannot answer it, and a client without it is refused.

​

generate_media

Generates images, videos or audio with any model of your workspace.

model · string · required

Model id, for example seedance-2.5.

mode · string · required

One of the model’s modes: text-to-image, image-to-video, reference-to-video…

input · object

The mode’s fields, as get_model lists them. A file field (first_frame, references, image…) takes an asset id, a URL or an element: see Your files.

max_credits · number

Refuse the generation, before spending, if it would cost more.

idempotency_key · string

A new UUID per generation, the same one on a retry: the retry returns the generation already started. Reusing a key for a different request is refused.

wait_seconds · integer · default: "40"

How long the call waits for the result, 0 to 40.

metadata · object

Up to 20 string pairs, to find the generation later with list_generations.

include_previews · boolean · default: "true"

Include small image previews in the result.

Image from a reference

{
  "model": "nano-banana-2.1",
  "mode": "image-to-image",
  "input": {
    "prompt": "Put the bottle on wet black stone with a rim light, same angle",
    "references": ["asset_7ab7bff259cb4b7e825f64ebff4595d7"]
  },
  "idempotency_key": "9d0c4e1a-2b7f-4c3e-8a5d-6f1b0e2c7a93"
}

The answer is the finished generation, or, past wait_seconds, its status and the next call:

Generation gen_6f1c2a9e4b7d4c1e9a3f0b5d8e2c7a14 (seedance-2.5 image-to-video) is processing.
Reserved 56 credits.
seedance-2.5 usually takes 1 min 30 s to 4 min once started.
Next: get_generation {"id":"gen_6f1c2a9e4b7d4c1e9a3f0b5d8e2c7a14","wait_seconds":40} until it is final.

​

get_generation

id · string · required

gen_…

wait_seconds · integer · default: "0"

Wait up to this many seconds (0 to 40) for the generation to finish.

A finished generation lists each output with its permanent URL and asset id, and what was charged and refunded.

​

add_asset

Exactly one source:

url · string

A public https file, downloaded by Argil. The same URL twice is the same asset.

upload · object

{ "filename", "content_type" }: returns a signed URL to PUT the file to, valid 10 minutes, for files up to 50 MB for an image, 2 GB for a video, 200 MB for an audio. Then get_asset with wait_seconds until it is ready.

file · object

{ "data_base64", "filename", "content_type" }: a file up to 2 MB, inline. For programs; never copy a file through a conversation.

​

Results

Every tool answers with a short text for the conversation, ending with Next: and the call to make when there is one, and the same data as structuredContent for programs. A refused call has isError: true and an error with a stable code: see Troubleshooting.

Tools keep their names and required arguments. New models, modes and optional arguments appear without breaking existing calls; a breaking change ships as a new tool, and the old one stays at least 12 months.