Skip to main content
Prompts are the core unit in Prompt Bank. This guide walks you through creating prompts for different media types, querying your prompt library with filters and pagination, making partial updates, and cleaning up prompts you no longer need. All examples use curl, but the same requests work with any HTTP client.

Create a prompt

1

Create a text prompt

Text prompts are useful for storing reusable LLM instructions or any content not tied to a specific image or video model. Send a POST request to /api/v1/prompts:
The API returns the full prompt object wrapped in the standard response envelope:
2

Create an image prompt

Image prompts target a specific model and can carry configuration like aspectRatio and referenceMediaUrls. The modelId field is required for image and video prompt types — it must match the pattern fal-ai/.... Supply a vaultId (integer) to add the prompt directly to an existing vault:

List prompts

Use GET /api/v1/prompts to retrieve your prompts. The endpoint supports cursor-based pagination and several filter parameters:

Fetch the first page

Fetch the next page

Pass the pagination.nextCursor value as the cursor query parameter to retrieve the following page:
When pagination.nextCursor is absent or null, you have reached the last page.

Filter by vault

Update a prompt

Use PATCH /api/v1/prompts/{promptId} to update one or more fields without replacing the entire object. Only include the fields you want to change:

Delete a prompt

Send a DELETE request to remove a prompt permanently. This action cannot be undone.
A successful deletion returns HTTP 200 with an empty data field:

Next steps