FluxPrompt API Reference

All API calls require your API key passed as a request header. Generate your key under Settings → API Keys.

⚠️ Never expose your API key in client-side code or public repositories. Rotate keys immediately if compromised.

Each agent's endpoint URL, email address, and input node IDs are unique to that flow. Open the Flow Global Node in your agent's settings to find:

⚠️ inputId values are unique to each flow — they cannot be reused across agents. Always copy them from the Flow Global Node.

The sessionId parameter is a core part of how FluxPrompt handles multi-user agents, chatbots, and any flow where memory needs to be isolated per user, client, or conversation.

When your agent uses a memory node, FluxPrompt stores and retrieves context scoped to the sessionId you pass. Every call made with the same sessionId shares the same memory thread — the agent can recall previous messages, extracted data, or any information stored during that session.

Without a sessionId, all calls to the agent share a single global memory context, which is rarely the right behaviour for multi-user applications.

Pass sessionId as a query parameter on every request where session continuity matters. The value can be any string — use whatever identifier is meaningful in your application.

Static session (e.g. a named client)

Dynamic session (e.g. per logged-in user)

Dynamic session per conversation

💡 A good pattern for chatbots is to generate a sessionId when a user starts a conversation and store it client-side for the duration of that chat thread.

POST https://api.fluxprompt.ai/flux/api-v2

bash

POST https://api.fluxprompt.ai/flux/api-v2-webhook

Same query parameters and JSON body as the REST API. Returns 200 OK on successful receipt.

Send an email to the agent's generated address. No API key required — authentication is handled via the unique address.

All API and Webhook calls use a JSON body with a single top-level field: variableInputs — an array of input objects, one per Data Input node you want to populate.

Minimal example

json

💡 Include multiple objects in variableInputs to populate several input nodes in a single call. Each inputId is unique to the flow — find yours in the Flow Global Node.

Each object in the variableInputs array targets a specific Data Input node.

Pass either links or files on a given input — not both.

Use links to pass URLs for the agent to retrieve and process.

Example

json

Use files to pass base64-encoded file content directly. isExtraction: true is required when using this field.

Example

json

Use typeOCR to control how attached files are read. Only applies when using files.

Memory Pods let you manage persistent document-based memory that agents can reference. The following endpoints allow you to add, replace, and delete documents within a pod.

💡 Your Pod ID is found in your FluxPrompt settings.

GET /flux/memory-pods/get

Returns a list of all memory pods available to your account.

POST /flux/memory-pods/{podId}/add-document

Upload a new document into a memory pod.

PUT /flux/memory-pods/{podId}/replace-document

Replace an existing document in a memory pod with a new one.

Request body

json

DELETE /flux/memory-pods/{podId}/delete-document

Remove a document from a memory pod.

A complete request body using all available fields across two input nodes.

json

Full request URL with session

ℹ️ The inputId values above are from a staging flow. Your flow will have different IDs — always copy them from your Flow Global Node.