This page has moved. Go to the Node Reference →

4.3 Node Reference

Every action inside a Flusso workflow is performed by a node. This reference documents every available node type: what it does, how to configure it, and what output it produces. Nodes are grouped by category, matching the sidebar organisation in the editor.

---

1. Triggers

Triggers are the starting point of every workflow. Only one trigger per workflow is allowed. The trigger fires based on an external event and passes its payload into the workflow context as {{ trigger.output.* }}.

Manual Trigger

Fires when you click the Run button in the editor or call the workflow via the API. Use for on-demand tasks and for testing during development.

Output: Any data passed as the run payload (via the API or test form).

Webhook Trigger

Fires when an HTTP POST request is received at the workflow's unique webhook URL (/api/webhooks/{workflow-id}). The request body, headers, and method are available in the output.

Output: payload (request body), headers, method.

Schedule Trigger

Fires on a repeating schedule defined by a cron expression. The workflow executes automatically at each scheduled time as long as it is in Active status.

Output: scheduled_at — the ISO timestamp when the run was triggered.

Event Trigger

Fires when another workflow emits a named event via the Emit Event node. Use this to chain workflows: one workflow emits, another listens.

Output: event_name, payload (data from the emitting workflow).

Error Trigger

Fires automatically when any other workflow in your account encounters a fatal error during execution. Use to build centralised error-handling or alerting workflows.

Output: workflow_id, run_id, step_key, error_message.

Platform-Specific Triggers

The following triggers work like the Webhook Trigger but validate platform-specific signatures:

• Stripe Trigger — validates Stripe webhook signatures.
• GitHub Trigger — validates GitHub webhook HMAC signatures.
• Shopify Trigger — validates Shopify HMAC-SHA256 webhook signatures.

Output: Same as Webhook Trigger — the request body is available as payload.

File Watch Triggers

These triggers fire when a new file is detected in a monitored location:

• Google Drive Watch — monitors a Google Drive folder.
• Dropbox Watch — monitors a Dropbox folder.
• Local File Watch — monitors a local directory on the server.

Output: file_name, file_path, file_size, mime_type.

---

2. AI & Agents

Agent

The most powerful node in Flusso. The Agent node sends a prompt to a large language model and returns its response. Agents can optionally perform RAG (retrieve relevant context from a knowledge base), call external tools, and maintain memory across runs.

Output: text (raw model response), json (parsed JSON if JSON Mode is on), usage (token counts).

Model Config Graph only

A configuration node (not executed) that specifies the AI model an Agent node uses in graph mode. Connect its output to the model_in port of an Agent node.

Memory Graph only

A configuration node that specifies a memory backend for an Agent node in graph mode. Connect to the memory_in port.

MCP Tool Graph only

A configuration node that defines an external tool accessible to an Agent node via the Model Context Protocol (MCP). Connect to the tool_in port. Multiple MCP Tool nodes can connect to the same Agent.

---

3. Data Retrieval & Processing

RAG (Knowledge Base Search)

Performs a semantic vector search against one of your Knowledge Bases. Returns the most relevant document excerpts, which you can then pass to an Agent node via its Context Sources setting.

Output: items — array of { text, score, metadata } objects.

Reranker

Re-ranks a list of search results using a cross-encoder model that evaluates each result's relevance to the query more precisely than vector similarity alone.

Output: items — re-ranked array of { text, score, metadata } objects.

Web Search

Performs a live web search and returns a list of results (title, URL, snippet). Useful for agents that need up-to-date information.

Output: results — array of { title, url, snippet }.

HTTP Request

Makes an outbound HTTP request to any URL. Use this node to interact with external REST APIs, fetch remote data, or send data to third-party services.

Output: status (HTTP status code), body (response body), headers.

Transform

Maps and reshapes data from a previous step into a new structure. Define a set of output fields, each with a name and a value expression (supports variable references and static values).

Output: An object with the defined fields as keys.

Schema Extractor

Uses an AI model to extract structured data from unstructured text (emails, documents, web pages). You define a JSON Schema describing the expected output; the model populates it.

Output: An object matching the defined schema.

Image Generation

Generates an image from a text prompt using an AI image model (e.g. DALL-E 3, Stable Diffusion).

Output: url (URL to the generated image), revised_prompt.

---

4. Flow Control

Logic Gate

Evaluates one or more conditions and routes the workflow to the true or false branch based on the result. Conditions are combined with AND or OR logic.

Output: result (true or false). Execution continues only down the matching branch.

Switch

Routes the workflow to one of several named branches based on the value of an expression. Similar to a switch/case statement in programming.

Output: matched_case — the case that matched.

Loop

Iterates over an array and executes the sub-steps for each item. The current item is available inside the loop as a variable.

Output: results — array of outputs from each iteration; count — number of iterations.

Parallel

Splits execution into multiple branches that run simultaneously. All branches must complete before the workflow continues past the parallel node.

Output: An object keyed by branch name, each containing the output of that branch's last step.

Stop

Explicitly terminates the workflow run with a given status. Use at the end of conditional branches to signal success or failure.

Wait

Pauses workflow execution for a fixed number of seconds before continuing.

Random

Generates a random number or selects a random item from a list. Useful for A/B testing, random routing, or simulation.

Output: value — the generated number or chosen item.

---

5. Data Manipulation

Set Variable

Creates or updates workflow variables that persist across steps. Unlike step outputs, variables can be updated multiple times during a workflow run.

Output: The set variables as key-value pairs.

Counter

Increments or decrements a numeric workflow variable. Useful inside loops to count items or track progress.

Output: value — the new value of the counter.

Filter

Filters an array to keep only items matching a condition.

Output: The filtered array stored in the specified output variable.

---

6. Notifications & Messaging

Email Notification

Sends an email using the SMTP server configured in Settings. Requires SMTP to be configured first.

Output: sent (boolean), message_id.

Webhook Notification

Sends an HTTP POST request to an external URL. Simpler than the HTTP Request node — intended for outbound notifications.

In-App Notification

Creates a notification visible in the Flusso notification bell for a specific user.

Slack Message

Sends a message to a Slack channel via the Slack API or a webhook URL.

Discord Message

Sends a message to a Discord channel via a Discord webhook URL.

Telegram Message

Sends a message to a Telegram chat via the Telegram Bot API.

Emit Event

Emits a named event that can trigger any workflow using an Event Trigger listening to the same event name. Use to chain workflows together.

Webhook Response

Sends an HTTP response back to the caller of a webhook-triggered workflow. Only applicable when the workflow is triggered via a webhook and the caller is waiting for a synchronous response.

---

7. File Operations

All file operation nodes share a common set of actions: read, write, and append.

Local File System

Reads or writes files on the local server filesystem.

Output: content (read), or path and bytes_written (write/append).

Google Drive

Reads or writes files in Google Drive. Requires a Google account connected in the workflow's Credentials drawer.

Dropbox

Reads or writes files in Dropbox. Configure Dropbox access credentials in the Integrations settings.

Amazon S3

Reads or writes files in an AWS S3 bucket. Configure your AWS credentials and bucket name.

Google Sheets

Reads or writes data in a Google Sheets spreadsheet. Requires Google integration authorised in the workflow's Credentials drawer.

Output: values (2D array for read), or updated_cells (for write).

---

8. Human / User Interaction

Approval

Pauses the workflow and waits for a human to approve or reject the continuation. The workflow enters Waiting status and the pending task appears on the Tasks page.

Behaviour: If approved, execution continues to the next step. If rejected, the workflow run is terminated with a failed status.

Manual Input

Pauses the workflow and waits for a human to fill in one or more form fields. The form appears on the Tasks page. When submitted, the entered values are available in the workflow's subsequent steps.

Output: An object with each field's name as the key and the user's input as the value.