Replicate

Overview

Replicate is a cloud AI platform that lets you run machine-learning models via API without setting up or managing servers and GPUs. With the Replicate integration in SquadOS, your agents can create predictions with any public model on the platform, fine-tune models, manage deployments and files, and orchestrate complete generative AI pipelines.

• Official website: https://replicate.com/
• Composio documentation: docs.composio.dev/toolkits/replicate

Authentication

This tool uses an API key (API_KEY) to connect.

You will need the following fields:

Field	Required	Description
api_key	Yes	API token generated in your Replicate account, used to authenticate all requests.

How to get credentials

1. Go to replicate.com and sign in (or create a free account).
2. Navigate to Account settings → API tokens (or go directly to replicate.com/account/api-tokens).
3. Click Create token, give the token a name, and confirm.
4. Copy the generated token — this is the value to enter in the api_key field when connecting in SquadOS.

Tip

Create a dedicated token for SquadOS. That way, you can revoke access without affecting other integrations that use different tokens.

How to connect in SquadOS

1. Go to Tools in the side menu (/admin/tools).
2. Open the Available tab and search for Replicate.
3. Click the card to open the details modal and hit Connect.
4. You’re taken to the secure connection page hosted by Composio, where you enter the API token obtained above.
5. Once done, you’re sent back to SquadOS with the account connected and the tool available for your agents. (Connection-flow details in Organization Tools.)

Available actions

Get Account Information

REPLICATE_ACCOUNT_GET

Tool to get authenticated account information. Use when you need to retrieve details about the account associated with the API token.

Output

Name	Type	Required	Description
data	string	Yes	Data from the action execution
error	string	No	Error if any occurred during the execution of the action
successful	boolean	Yes	Whether or not the action execution was successful or not

Cancel Prediction

REPLICATE_CANCEL_PREDICTION

Tool to cancel a prediction that is still running. Use when you need to stop an in-progress prediction to free up resources or halt execution.

Input parameters

Name	Type	Required	Description
prediction_id	string	Yes	The ID of the prediction to cancel

Output

Name	Type	Required	Description
data	string	Yes	Data from the action execution
error	string	No	Error if any occurred during the execution of the action
successful	boolean	Yes	Whether or not the action execution was successful or not

Get model collection

REPLICATE_COLLECTIONS_GET

Tool to get a specific collection of models by its slug. Use when you need detailed information about a collection and its models.

Input parameters

Name	Type	Required	Description
collection_slug	string	Yes	The slug identifier of the collection, e.g., ‘super-resolution’ or ‘text-to-image’

Output

Name	Type	Required	Description
data	string	Yes	Data from the action execution
error	string	No	Error if any occurred during the execution of the action
successful	boolean	Yes	Whether or not the action execution was successful or not

List model collections

REPLICATE_COLLECTIONS_LIST

Tool to list all collections of models. Use when you need to retrieve available model collections. Collections are curated groupings of related models. Response includes only collection metadata (name, slug, description), not individual models within each collection; use REPLICATE_MODELS_GET for per-model details. Response may include a non-null next field indicating additional pages; follow it to enumerate all collections.

Input parameters

Name	Type	Required	Description
cursor	string	No	Cursor for pagination. Use the cursor from ‘next’ URL of previous response.

Output

Name	Type	Required	Description
data	string	Yes	Data from the action execution
error	string	No	Error if any occurred during the execution of the action
successful	boolean	Yes	Whether or not the action execution was successful or not

Create Model

REPLICATE_CREATE_MODEL

Tool to create a new Replicate model with specified owner, name, visibility, and hardware. Use when you need to create a destination model before launching LoRA/fine-tune training.

Input parameters

Name	Type	Required	Description
name	string	Yes	The name of the model. This must be unique among all models owned by the user or organization.
owner	string	Yes	The name of the user or organization that will own the model. This must be the same as the user or organization making the API request.
hardware	string	Yes	The SKU for the hardware used to run the model. Possible values can be retrieved from the hardware.list endpoint.
paper_url	string	No	A URL for the model’s paper
github_url	string	No	A URL for the model’s source code on GitHub
visibility	string	Yes	Whether the model should be public or private. Public models are viewable and runnable by anyone; private models are restricted to owners.
description	string	No	A description of the model
license_url	string	No	A URL for the model’s license
cover_image_url	string	No	A URL for the model’s cover image. This should be an image file.

Output

Name	Type	Required	Description
data	string	Yes	Data from the action execution
error	string	No	Error if any occurred during the execution of the action
successful	boolean	Yes	Whether or not the action execution was successful or not

Create Prediction (Deployment)

REPLICATE_CREATE_PREDICTION

Tool to create a prediction for a Replicate Deployment. IMPORTANT: This action ONLY works with Replicate Deployments (persistent instances you create and manage), NOT public models. Deployments are created via REPLICATE_DEPLOYMENTS_CREATE. To run public models (e.g., ‘meta/llama-2-70b-chat’, ‘stability-ai/sdxl’), use REPLICATE_MODELS_PREDICTIONS_CREATE instead. Use ‘wait_for’ to wait until the prediction completes.

Input parameters

Name	Type	Required	Description
input	object	Yes	JSON object of inputs for the model. Keys and types must exactly match the deployment model’s schema; consult REPLICATE_MODELS_README_GET or REPLICATE_MODELS_EXAMPLES_LIST first. For file inputs, verify file IDs via REPLICATE_FILES_LIST before submitting.
webhook	string	No	HTTPS URL to receive webhook callbacks
wait_for	integer	No	Seconds to wait for prediction completion; adds a ‘Prefer: wait=<n>’ header if provided. Valid range is 1–60; if the prediction does not complete within this window, poll for status or use a webhook instead.
deployment_name	string	Yes	Name of the deployment. NOTE: This is the deployment name, NOT a public model name. Deployments are created via REPLICATE_DEPLOYMENTS_CREATE. To run public models, use REPLICATE_MODELS_PREDICTIONS_CREATE instead.
deployment_owner	string	Yes	Username or organization that owns the deployment. NOTE: This is for Replicate Deployments only, NOT public models. To run public models, use REPLICATE_MODELS_PREDICTIONS_CREATE instead.
webhook_events_filter	array	No	List of events to trigger webhook calls

Output

Name	Type	Required	Description
data	string	Yes	Data from the action execution
error	string	No	Error if any occurred during the execution of the action
successful	boolean	Yes	Whether or not the action execution was successful or not

Create Deployment

REPLICATE_DEPLOYMENTS_CREATE

Tool to create a new deployment with specified model, version, hardware, and scaling parameters. Use when you need to deploy a model for production use with auto-scaling.

Input parameters

Name	Type	Required	Description
name	string	Yes	The deployment’s identifier
model	string	Yes	Full model name in format owner/name
version	string	Yes	The 64-character model version ID to deploy
hardware	string	Yes	Hardware SKU from the hardware.list endpoint
max_instances	integer	Yes	Maximum instances for auto-scaling
min_instances	integer	Yes	Minimum instances for auto-scaling

Output

Name	Type	Required	Description
data	string	Yes	Data from the action execution
error	string	No	Error if any occurred during the execution of the action
successful	boolean	Yes	Whether or not the action execution was successful or not

Delete Deployment

REPLICATE_DEPLOYMENTS_DELETE

Tool to delete a deployment from your account. Use when you need to remove a deployment. Deployments must be offline and unused for at least 15 minutes before deletion.

Input parameters

Name	Type	Required	Description
deployment_name	string	Yes	The name of the deployment to delete. Deployment must be offline and unused for at least 15 minutes before deletion.
deployment_owner	string	Yes	The username or organization that owns the deployment

Output

Name	Type	Required	Description
data	string	Yes	Data from the action execution
error	string	No	Error if any occurred during the execution of the action
successful	boolean	Yes	Whether or not the action execution was successful or not

Get Deployment Details

REPLICATE_DEPLOYMENTS_GET

Tool to get deployment details by owner and name. Use when you need information about a specific deployment including its release configuration and hardware settings.

Input parameters

Name	Type	Required	Description
deployment_name	string	Yes	The name of the deployment
deployment_owner	string	Yes	The name of the user or organization that owns the deployment

Output

Name	Type	Required	Description
data	string	Yes	Data from the action execution
error	string	No	Error if any occurred during the execution of the action
successful	boolean	Yes	Whether or not the action execution was successful or not

List Deployments

REPLICATE_DEPLOYMENTS_LIST

Tool to list all deployments associated with the account. Use when you need to retrieve deployment configurations and their latest releases.

Input parameters

Name	Type	Required	Description
cursor	string	No	Cursor for pagination. Use the cursor from ‘next’ URL of previous response.

Output

Name	Type	Required	Description
data	string	Yes	Data from the action execution
error	string	No	Error if any occurred during the execution of the action
successful	boolean	Yes	Whether or not the action execution was successful or not

Create File

REPLICATE_FILES_CREATE

Tool to create or upload a file to Replicate. Use when you need to upload file content with optional metadata.

Input parameters

Name	Type	Required	Description
content	object	Yes	File to upload.
metadata	object	No	User-provided metadata associated with the file (must be valid JSON)
content_type	string	No	The content/MIME type for the file

Output

Name	Type	Required	Description
data	string	Yes	Data from the action execution
error	string	No	Error if any occurred during the execution of the action
successful	boolean	Yes	Whether or not the action execution was successful or not

Delete File

REPLICATE_FILES_DELETE

Tool to delete a file by its ID. Use when you need to remove a file from storage. Returns 204 No Content on success.

Input parameters

Name	Type	Required	Description
file_id	string	Yes	The unique identifier of the file to delete

Output

Name	Type	Required	Description
data	string	Yes	Data from the action execution
error	string	No	Error if any occurred during the execution of the action
successful	boolean	Yes	Whether or not the action execution was successful or not

Get File Details

REPLICATE_FILES_GET

Tool to get details of a file by its ID. Use when you need to inspect uploaded file information before further operations. Returned URLs may be short-lived; download or persist needed files promptly after retrieval.

Input parameters

Name	Type	Required	Description
file_id	string	Yes	The ID of the file to retrieve. Must originate from a prior upload or prediction output. Use REPLICATE_FILES_LIST to confirm the ID exists before calling this tool.

Output

Name	Type	Required	Description
data	string	Yes	Data from the action execution
error	string	No	Error if any occurred during the execution of the action
successful	boolean	Yes	Whether or not the action execution was successful or not

List Files

REPLICATE_FILES_LIST

Tool to retrieve a paginated list of uploaded files. Use to view all files created by the authenticated user or organization. Files are sorted with most recent first. Pagination is cursor-based: follow the next cursor until empty to retrieve all files. Limit requests to 1–2/second to avoid 429 Too Many Requests errors. Use to validate current file_ids before passing to prediction tools, as stale file_ids cause runtime errors.

Input parameters

Name	Type	Required	Description
cursor	string	No	Cursor for pagination. Use the cursor from ‘next’ URL of previous response.

Output

Name	Type	Required	Description
data	string	Yes	Data from the action execution
error	string	No	Error if any occurred during the execution of the action
successful	boolean	Yes	Whether or not the action execution was successful or not

Get Prediction

REPLICATE_GET_PREDICTION

Tool to get the status and output of a prediction by its ID. Use when you need to check on a running prediction or retrieve the results of a completed prediction.

Input parameters

Name	Type	Required	Description
prediction_id	string	Yes	The ID of the prediction to get

Output

Name	Type	Required	Description
data	string	Yes	Data from the action execution
error	string	No	Error if any occurred during the execution of the action
successful	boolean	Yes	Whether or not the action execution was successful or not

List Available Hardware

REPLICATE_HARDWARE_LIST

Tool to list available hardware SKUs for models and deployments. Use when you need to see what hardware options are available on the Replicate platform.

Output

Name	Type	Required	Description
data	string	Yes	Data from the action execution
error	string	No	Error if any occurred during the execution of the action
successful	boolean	Yes	Whether or not the action execution was successful or not

List model examples

REPLICATE_MODELS_EXAMPLES_LIST

Tool to list example predictions for a specific model. Use when you want to retrieve author-provided illustrative examples after identifying the model. Returned examples are minimal working payloads; cross-reference with REPLICATE_MODELS_README_GET before calling REPLICATE_CREATE_PREDICTION to satisfy strict input validation.

Input parameters

Name	Type	Required	Description
cursor	string	No	Cursor for pagination. Use the cursor from ‘next’ URL of previous response.
model_name	string	Yes	The name of the model to list examples for. Must exactly match the Replicate URL slug (case-sensitive).
model_owner	string	Yes	The name of the user or organization that owns the model. Must exactly match the Replicate URL slug (case-sensitive).

Output

Name	Type	Required	Description
data	string	Yes	Data from the action execution
error	string	No	Error if any occurred during the execution of the action
successful	boolean	Yes	Whether or not the action execution was successful or not

Get Model Details

REPLICATE_MODELS_GET

Tool to get details of a specific model by owner and name. Consult the returned input schema before constructing any prediction request — each model defines its own required/optional fields (e.g., prompt, aspect_ratio, version); missing or unknown keys cause validation errors. Model schemas and available versions may change over time; recheck before production use.

Input parameters

Name	Type	Required	Description
model_name	string	Yes	The name of the model, e.g., ‘hello-world’. Must be an exact, case-sensitive Replicate slug (lowercase, hyphenated); e.g., use ‘hello-world’ not ‘Hello World’.
model_owner	string	Yes	The username or organization that owns the model, e.g., ‘replicate’. Must be an exact, case-sensitive Replicate slug (lowercase, hyphenated); e.g., use ‘replicate’ not ‘Replicate’.

Output

Name	Type	Required	Description
data	string	Yes	Data from the action execution
error	string	No	Error if any occurred during the execution of the action
successful	boolean	Yes	Whether or not the action execution was successful or not

List Public Models

REPLICATE_MODELS_LIST

Tool to list public models with pagination and sorting. Use when you need to browse available models or find models sorted by creation date.

Input parameters

Name	Type	Required	Description
cursor	string	No	Pagination cursor for navigating through results. Use the ‘next’ URL from previous responses to fetch the next page of results.
sort_by	string (model_created_at | latest_version_created_at)	No	Field to sort models by. Options are ‘model_created_at’ (when model was created) or ‘latest_version_created_at’ (when latest version was created). Defaults to ‘latest_version_created_at’ if not specified.
sort_direction	string (asc | desc)	No	Sort direction - ‘asc’ for ascending or ‘desc’ for descending. Defaults to ‘desc’ (newest first) if not specified.

Output

Name	Type	Required	Description
data	string	Yes	Data from the action execution
error	string	No	Error if any occurred during the execution of the action
successful	boolean	Yes	Whether or not the action execution was successful or not

Create Model Prediction

REPLICATE_MODELS_PREDICTIONS_CREATE

Tool to create a prediction using an official Replicate model. Use when you need to run inference with a specific model using its owner and name. Supports synchronous waiting (up to 60 seconds) and webhooks for async notifications.

Input parameters

Name	Type	Required	Description
input	object	Yes	Model inputs as JSON object. Schema varies by model - each model has its own set of required fields, optional parameters, and validation constraints (e.g., max dimensions for image models). IMPORTANT: Before making predictions, use the REPLICATE_MODELS_GET action to fetch the model’s openapi_schema which contains all valid input parameters and their constraints. For image models like flux-1.1-pro, common constraints include max width/height limits (e.g., 1440px), valid aspect ratios, and output format options.
webhook	string	No	HTTPS URL to receive POST requests when prediction updates occur
wait_for	integer	No	Seconds to wait for prediction completion (1-60); adds a ‘Prefer: wait=<n>’ header to block and wait synchronously
model_name	string	Yes	The name of the model as it appears on Replicate. Must be an exact match. Find valid model names by searching at replicate.com or using the search/models_list actions. The full model identifier is {model_owner}/{model_name}.
model_owner	string	Yes	The username or organization that owns the model on Replicate. Find valid owners by searching models at replicate.com or using the search/models_list actions. Examples: ‘meta’, ‘stability-ai’, ‘black-forest-labs’, ‘bytedance’, ‘google’.
cancel_after	string	No	Maximum execution duration with optional unit suffixes (s/m/h); minimum 5 seconds. Example: ’60s’ or ‘5m’.
webhook_events_filter	array	No	Filter webhook triggers by event type: ‘start’, ‘output’, ‘logs’, or ‘completed’

Output

Name	Type	Required	Description
data	string	Yes	Data from the action execution
error	string	No	Error if any occurred during the execution of the action
successful	boolean	Yes	Whether or not the action execution was successful or not

Get Model README

REPLICATE_MODELS_README_GET

Tool to get the README content for a model in Markdown format. Consult alongside REPLICATE_MODELS_EXAMPLES_LIST before calling REPLICATE_CREATE_PREDICTION — Replicate enforces strict JSON schemas on model inputs and returns 422 errors for incorrect keys or types. Use after retrieving model details when you want to view its documentation.

Input parameters

Name	Type	Required	Description
model_name	string	Yes	The name of the model, e.g., ‘hello-world’
model_owner	string	Yes	The username or organization that owns the model, e.g., ‘replicate’. Must exactly match the Replicate URL slug (case-sensitive).

Output

Name	Type	Required	Description
data	string	Yes	Data from the action execution
error	string	No	Error if any occurred during the execution of the action
successful	boolean	Yes	Whether or not the action execution was successful or not

Get Model Version

REPLICATE_MODELS_VERSIONS_GET

Tool to get a specific version of a model. Use when you need details about a particular model version including its schema and metadata.

Input parameters

Name	Type	Required	Description
model_name	string	Yes	The name of the model, e.g., ‘hello-world’
version_id	string	Yes	The 64-character identifier of the version
model_owner	string	Yes	The name of the user or organization that owns the model, e.g., ‘replicate’

Output

Name	Type	Required	Description
data	string	Yes	Data from the action execution
error	string	No	Error if any occurred during the execution of the action
successful	boolean	Yes	Whether or not the action execution was successful or not

List Model Versions

REPLICATE_MODELS_VERSIONS_LIST

Tool to list all versions of a specific model. Use when you need to see all available versions of a model, sorted by newest first.

Input parameters

Name	Type	Required	Description
cursor	string	No	Cursor for pagination. Use the cursor from ‘next’ URL of previous response.
model_name	string	Yes	The name of the model, e.g., ‘hello-world’
model_owner	string	Yes	The username or organization that owns the model, e.g., ‘replicate’

Output

Name	Type	Required	Description
data	string	Yes	Data from the action execution
error	string	No	Error if any occurred during the execution of the action
successful	boolean	Yes	Whether or not the action execution was successful or not

Create Prediction by Version

REPLICATE_PREDICTIONS_CREATE

Tool to create a prediction to run a model by version ID. Use when you have a specific model version identifier and need to run inference with provided inputs. Supports synchronous waiting and webhook notifications.

Input parameters

Name	Type	Required	Description
input	object	Yes	Model inputs as JSON object. Schema varies by model. Files should be HTTP URLs (for files >256kb or reusable) or data URLs (for files ≤256kb)
stream	boolean	No	Deprecated. Streaming availability is now included in response urls
version	string	Yes	Model version identifier. Accepts formats: {owner}/{name}, {owner}/{name}:{version_id}, or {version_id} alone
webhook	string	No	HTTPS URL to receive POST requests when prediction updates occur. Body mirrors the get prediction response
wait_for	integer	No	Seconds (1-60) to hold connection open until model finishes. Adds ‘Prefer: wait=<n>’ header
cancel_after	string	No	Maximum runtime before automatic cancellation. Accepts durations with optional suffixes (s/m/h). Minimum: 5 seconds. Examples: ’30s’, ‘5m’, ‘1h30m45s’
webhook_events_filter	array	No	Controls webhook triggers. Options: ‘start’, ‘output’, ‘logs’, ‘completed’. Output/logs events throttled to once per 500ms

Output

Name	Type	Required	Description
data	string	Yes	Data from the action execution
error	string	No	Error if any occurred during the execution of the action
successful	boolean	Yes	Whether or not the action execution was successful or not

List All Predictions

REPLICATE_PREDICTIONS_LIST

Tool to list all predictions for the authenticated user or organization with pagination. Use when you need to retrieve prediction history or filter predictions by creation date.

Input parameters

Name	Type	Required	Description
cursor	string	No	Cursor for pagination. Use the cursor from ‘next’ URL of previous response.
created_after	string	No	Include only predictions created at or after this date-time, in ISO 8601 format (e.g., ‘2024-01-01T00:00:00Z’). Useful for filtering recent predictions.
created_before	string	No	Include only predictions created before this date-time, in ISO 8601 format (e.g., ‘2024-12-31T23:59:59Z’). Useful for filtering older predictions.

Output

Name	Type	Required	Description
data	string	Yes	Data from the action execution
error	string	No	Error if any occurred during the execution of the action
successful	boolean	Yes	Whether or not the action execution was successful or not

Search Models and Collections

REPLICATE_SEARCH

Tool to search for models, collections, and docs using text queries (beta). Use when you need to find relevant models or collections based on keywords or descriptions.

Input parameters

Name	Type	Required	Description
limit	integer	No	Maximum number of model results to return. Must be between 1 and 50. Defaults to 20 if not specified.
query	string	Yes	The search query string to execute. Can search across models, collections, and documentation.

Output

Name	Type	Required	Description
data	string	Yes	Data from the action execution
error	string	No	Error if any occurred during the execution of the action
successful	boolean	Yes	Whether or not the action execution was successful or not

Cancel Training

REPLICATE_TRAININGS_CANCEL

Tool to cancel an ongoing training operation in Replicate. Use when you need to stop a training job that is in progress.

Input parameters

Name	Type	Required	Description
training_id	string	Yes	The identifier of the training operation to be terminated

Output

Name	Type	Required	Description
data	string	Yes	Data from the action execution
error	string	No	Error if any occurred during the execution of the action
successful	boolean	Yes	Whether or not the action execution was successful or not

Create Training Job

REPLICATE_TRAININGS_CREATE

Tool to create a training job for a specific model version. Use when you need to fine-tune a model with custom training data. Supports webhook notifications for training status updates.

Input parameters

Name	Type	Required	Description
name	string	Yes	Name of the model to train
input	object	Yes	JSON object containing inputs to the Cog model’s train() function. Schema varies by model
owner	string	Yes	Name of the user or organization owning the model
webhook	string	No	HTTPS URL to receive POST requests when training updates occur
version_id	string	Yes	The 64-character model version identifier to use for training
destination	string	Yes	Destination model to push to in format {owner}/{name}
webhook_events_filter	array	No	List of event types that trigger webhook calls. Options: ‘start’, ‘output’, ‘logs’, ‘completed’

Output

Name	Type	Required	Description
data	string	Yes	Data from the action execution
error	string	No	Error if any occurred during the execution of the action
successful	boolean	Yes	Whether or not the action execution was successful or not

List Training Jobs

REPLICATE_TRAININGS_LIST

Tool to list all training jobs for the authenticated user or organization with pagination. Use when you need to retrieve training history or check the status of training jobs.

Input parameters

Name	Type	Required	Description
cursor	string	No	Cursor for pagination. Use the cursor from ‘next’ URL of previous response.

Output

Name	Type	Required	Description
data	string	Yes	Data from the action execution
error	string	No	Error if any occurred during the execution of the action
successful	boolean	Yes	Whether or not the action execution was successful or not

Update Model Metadata

REPLICATE_UPDATE_MODELS

Tool to update metadata for a model including description, URLs, and README. Use when you need to modify a model’s visibility, documentation, or associated links.

Input parameters

Name	Type	Required	Description
readme	string	No	The README content of the model
paper_url	string	No	A URL for the model’s paper
github_url	string	No	A URL for the model’s source code on GitHub
model_name	string	Yes	The name of the model to update
description	string	No	A description of the model
license_url	string	No	A URL for the model’s license
model_owner	string	Yes	The name of the user or organization that owns the model
weights_url	string	No	A URL for the model’s weights

Output

Name	Type	Required	Description
data	string	Yes	Data from the action execution
error	string	No	Error if any occurred during the execution of the action
successful	boolean	Yes	Whether or not the action execution was successful or not

Get Webhook Signing Secret

REPLICATE_WEBHOOKS_SECRET_GET

Tool to get the signing secret for the default webhook. Use when you need to retrieve the secret key used to verify webhook authenticity.

Output

Name	Type	Required	Description
data	string	Yes	Data from the action execution
error	string	No	Error if any occurred during the execution of the action
successful	boolean	Yes	Whether or not the action execution was successful or not