Skip to main content
Actions are the core building blocks of ServFlow workflows. They perform discrete operations such as fetching data from a database, calling external APIs, running AI agents, or transforming values. Every workflow consists of an entry point, one or more actions, and a response. You can configure actions in two ways:
  • Dashboard UI — Visual interface for building workflows by selecting action types and filling in configuration fields. Best for learning, prototyping, and rapid development.
  • YAML Configuration — Code-based configuration files for version control, CI/CD pipelines, and production deployments.
Both approaches produce the same result — the dashboard saves configurations as YAML files, so you can seamlessly switch between them. See Running Modes to understand when to use each approach.

​
Working with Actions in the Dashboard

The ServFlow dashboard provides a visual interface for adding and configuring actions in your workflows.

​
Adding an Action

To add an action to your workflow:
  1. Click the + button below an existing node in the workflow canvas
  2. Select Actions from the node type options
  3. Browse the list of available actions in the sidebar

​
Browsing Available Actions

When you click to add an action, a sidebar panel displays all available action types. Each action shows its name and a brief description of what it does.
Available actions list in the ServFlow dashboard sidebar

​
Configuring an Action

Selecting an action opens the configuration panel on the left side of the screen. Each action type has specific fields to configure.
Action configuration panel showing AI Agent settings
The configuration panel displays:
  • Action type — The selected action (e.g., AI Agent)
  • Configuration fields — Type-specific settings such as identifiers, prompts, or connection details
  • Save Action — Button to save your configuration
Use {{ .Variable }} syntax in input fields to reference dynamic content from previous actions or request parameters. See Dynamic Content for more details.

​
Action Outputs

Actions have output handles that let you define the flow:
  • Next — Where to route on successful completion
  • Fail — Where to route when an error occurs
  • Tools — Additional outputs for specific action types (e.g., AI Agent tools)
Drag connections from these handles to other actions, conditionals, or response nodes to build your workflow logic.

​
YAML Configuration

Actions can also be defined directly in YAML configuration files. This approach is preferred for:
  • Version control — Track API changes in Git
  • CI/CD pipelines — Deploy configurations automatically
  • Production deployments — Run in Headless Mode without the dashboard
  • Programmatic generation — Generate configs from code or templates
See the YAML Quickstart for a complete tutorial on building workflows with YAML.

​
Action Structure

Every action follows this structure in YAML: 
actions:
  action_id:
    type: action_type
    config:
      # type-specific configuration
    next: action.next_step
    fail: response.error
The action_id is a unique identifier you choose for the action. You’ll use this ID to reference the action’s results in other parts of your workflow.

​
Common Fields

FieldTypeRequiredDescription
typestringYesThe action type identifier (e.g., fetch, http, agent)
configobjectNoType-specific configuration options
nextstringNoStep to execute on success
failstringNoStep to execute on failure

​
Chaining Actions

Use the next and fail fields to define your workflow’s execution path:
  • next — Routes to another step when the action succeeds
  • fail — Routes to an error handler when the action fails
The routing syntax follows the pattern type.identifier:
  • action.other_action — Route to another action
  • conditional.check_value — Route to a conditional branch
  • response.success — Route to a response
Here’s an example of chaining actions where the second action uses data from the first: 
actions:
  get_user:
    type: fetch
    config:
      integrationID: my_database
      table: users
      filters:
        - field: id
          operator: eq
          value: "{{ param \"user_id\" }}"
      single: true
    next: action.send_welcome

  send_welcome:
    type: email
    config:
      recipientEmail: "{{ .get_user.email }}"
      subject: "Welcome!"
      content: "Hello {{ .get_user.name }}, welcome to our platform."
    next: response.success

​
Accessing Action Results

When an action completes, its result is stored and can be referenced by subsequent nodes. Use the {{ .action_id }} syntax to access these values. 
# Access the full result from action 'fetch_user'
value: "{{ .fetch_user }}"

# Access a nested field
value: "{{ .fetch_user.email }}"

# Use in expressions
expression: "{{ gt (len .fetch_user) 0 }}"
For example, if an action with ID get_user returns user data, you can access the email field in a later action with {{ .get_user.email }}.

​
Filter Syntax

Several actions use filters to query data. Filters follow this structure: 
filters:
  - field: field_name
    operator: eq
    value: "comparison_value"

​
Available Operators

OperatorDescription
eqEqual to
neNot equal to
gtGreater than
gteGreater than or equal
ltLess than
lteLess than or equal
inValue in array
ninValue not in array
Example with multiple filters: 
filters:
  - field: status
    operator: eq
    value: active
  - field: age
    operator: gte
    value: "18"
  - field: role
    operator: in
    value:
      - admin
      - moderator

​
Available Actions

ServFlow provides actions organized by category:

​
Data Operations

ActionDescription
fetchRetrieves data from a database table
storeStores a new record in a database table
updateUpdates existing records
deleteDeletes records from a database table
mongoqueryExecutes native MongoDB queries

​
Vector Operations

ActionDescription
storevectorStores vector embeddings
fetchvectorsQueries vectors for similarity search

​
HTTP & External APIs

ActionDescription
httpMakes HTTP requests to external APIs

​
Authentication & Security

ActionDescription
authenticateValidates JWT tokens and fetches user data
jwtEncodes or decodes JWT tokens
hashHashes values using bcrypt

​
Transformation & Logic

ActionDescription
javascriptExecutes custom JavaScript code
staticReturns a static or computed value

​
AI & Agents

ActionDescription
agentInteracts with AI models

​
Flow Control

ActionDescription
parallelExecutes multiple actions simultaneously

​
Communication

ActionDescription
emailSends email messages via SMTP

​
System

ActionDescription
commandExecutes shell commands

​
Memory

ActionDescription
memory_storeStores data in temporary memory
memory_fetchRetrieves data from temporary memory

​
Binance Trading

ActionDescription
binance/getpriceGets cryptocurrency price data
binance/accountbalanceQueries Binance account balances
binance/spotorderPlaces spot trading orders
binance/futuresorderPlaces futures orders
binance/pricedifferenceCalculates price changes
binance/tradeinfoGets order and position info

​
Next Steps

Data Operations

Learn how to fetch, store, update, and delete data.

HTTP Requests

Call external APIs from your workflows.

AI Agents

Integrate AI models into your workflows.

Conditionals

Add branching logic to your workflows.