Input / Output
==============

The **Input** and **Output** components are required for every Agentic Flow. The Playground is only available when both are present.

----

.. _component-ref-input:

Input
-----

The entry point for the flow. Receives the user's message or trigger payload and passes it downstream.

**Use in a flow**

1. Add an **Input** node to the canvas (typically at the start of the flow).
2. Connect the **Message** output to downstream components (for example, an Agent, Prompt, or Tool Calling Agent).
3. Optionally connect **Conversation History** if a downstream step needs prior messages.

.. code-block:: text

   Input (Message) ──► Agent / Prompt / Processing ──► Output

In the Playground, users type messages in the **Text** field. When the flow runs via API or channels, the message or trigger payload passes through the same output.

**Parameters**

.. list-table::
   :widths: 30 10 10 50
   :header-rows: 1

   * - Parameter
     - Default
     - Hidden
     - Description
   * - Text
     - —
     - No
     - The message to pass as input. Used in the Playground and when no external input is wired.
   * - Conversation History
     - —
     - No
     - Prior conversation messages to include as context.
   * - Store Messages
     - true
     - Yes
     - Whether to store the message in chat history.
   * - Sender Type
     - User
     - Yes
     - Identifies the sender as User or Language Model.
   * - Sender Name
     - User
     - Yes
     - Display name for the sender.
   * - Session ID
     - —
     - Yes
     - Chat session identifier. Uses the current session if empty.
   * - Files
     - —
     - Yes
     - Files to attach with the message (PDF, DOCX, Excel, PPTX, images, text).
   * - Background Color
     - —
     - Yes
     - Background color for the message icon in chat.
   * - Icon
     - —
     - Yes
     - Icon for the message in chat.
   * - Text Color
     - —
     - Yes
     - Text color for the sender name in chat.

**Outputs**

.. list-table::
   :widths: 25 75
   :header-rows: 1

   * - Output
     - Description
   * - Message
     - The user's input as a message. Connect to downstream components.
   * - Conversation History
     - Conversation history as a message. Use when downstream steps need chat context.

----

.. _component-ref-output:

Output
------

The exit point of the flow. Returns the final result to the caller or connected channel.

**Use in a flow**

1. Add an **Output** node to the canvas (typically at the end of the flow).
2. Connect the final response to **Text** (from an Agent, LLM, or other component that produces a message).
3. Run the flow in the Playground or via API to see the result.

.. code-block:: text

   Agent / LLM ──► Output (Text)
   Input ──► ... ──► Output

**Parameters**

.. list-table::
   :widths: 30 10 10 50
   :header-rows: 1

   * - Parameter
     - Default
     - Hidden
     - Description
   * - Text
     - —
     - No
     - The bot response to display. Accepts Message, Data, or DataFrame input.
   * - Store Messages
     - true
     - Yes
     - Whether to store the message in chat history.
   * - Sender Type
     - Language Model
     - Yes
     - Identifies the sender as User or Language Model.
   * - Sender Name
     - AI
     - Yes
     - Display name for the sender.
   * - Session ID
     - —
     - Yes
     - Chat session identifier.
   * - Background Color
     - —
     - Yes
     - Background color for the message icon.
   * - Icon
     - —
     - Yes
     - Icon for the message.
   * - Text Color
     - —
     - Yes
     - Text color for the sender name.
   * - Channel Delivery
     - false
     - No
     - When enabled, delivers the message to configured outbound channels.
   * - Assume Policy
     - —
     - No
     - Access policy for channel delivery and resource access.

When **Channel Delivery** is enabled, open **Controls** to configure delivery targets, routing, and LLM settings for outbound channels.

**Outputs**

.. list-table::
   :widths: 25 75
   :header-rows: 1

   * - Output
     - Description
   * - Message
     - The formatted response message returned by the flow.