Create and Edit Pipelines from Canvas

Create, configure, and manage pipelines directly from Canvas while staying in the conversation flow.

Pipeline Canvas Feature: Visual Pipeline Management

The Pipeline Canvas interface in ELITEA serves as an integrated pipeline management system accessible directly from the chat interface. This feature enables you to create, configure, and manage AI pipelines without leaving your conversation context, streamlining your development workflow.

Integrated Chat Experience

Access pipeline management directly from the PARTICIPANTS section in chat, maintaining conversation context while managing AI workflows.

Dual-Mode Configuration

Use both Configuration mode for general settings and Flow Editor mode for visual workflow design.

Real-time Validation

Configuration fields are validated in real-time, ensuring proper setup before pipeline creation.

Instant Integration

Created pipelines are immediately available for use in conversations and can be added to the PARTICIPANTS section.

Advanced Model Configuration

Integration with ELITEA’s LLM model management system for comprehensive AI model selection and fine-tuning.

---

Creating Pipelines via Canvas Interface

1. Access the Pipeline Creation Canvas

   1. Navigate to the Chat page (main sidebar menu).

   2. In the PARTICIPANTS section, locate Pipelines.

   3. Click on the Create new pipeline button.

      

   The “Create New Pipeline” canvas interface will be displayed with all available configuration sections.

2. Configure General Information

   The pipeline creation form opens in a single-page layout. Fill in the GENERAL section:

   • Name* (required): Enter a unique, descriptive name for your pipeline (e.g., “Content Review Workflow”, “Data Processing Pipeline”)
   • Description* (required): Provide a clear description of what your pipeline will do and its purpose
   • Tags (optional): Add relevant tags by typing tag names to help organize and categorize your pipelines

   Required fields are marked with an asterisk * and must be completed before the pipeline can be created.

   Welcome Message (Optional)

   In the WELCOME MESSAGE section, add a message that users see when they first interact with your pipeline:

   • Example: “Hello! I’m your content review pipeline. I’ll help you process and validate your content through multiple review stages.”

   Conversation Starters (Optional)

   In the CONVERSATION STARTERS section:

   1. Click the + Starter button to add conversation starters (maximum 4 starters allowed)
   2. Enter helpful prompts that guide users on how to interact with your pipeline
   3. Each starter has a delete (trash) icon for removal if needed
   4. Examples:
      • “Start a new review process”

   Advanced Settings (Optional)

   In the ADVANCED section:

   • Step limit: Set the maximum number of steps the pipeline can execute (default: 25). This controls how many operations the pipeline can perform in a single execution.

   LLM Model

   The LLM model selector is available during creation. Click Select LLM Model to choose the model your pipeline will use, and click the settings icon (⚙️) to configure model parameters.

   
3. Save Initial Configuration

   Click the Save button to create your pipeline. The canvas transitions to the advanced configuration interface with two tabs:

   • Configuration — Add toolkits, agents, pipelines, and MCP connections (these buttons are disabled until the pipeline is saved).

   • Flow editor — Visually design the pipeline workflow using nodes and connectors.

     

---

Advanced Pipeline Configuration

After saving the initial configuration, the advanced configuration interface opens with two main tabs:

Configuration Tab

LLM Model and Settings Configuration

1. Model Selection:

   • Click the Select LLM Model button at the top of the interface
   • Choose from available LLM models in your project (e.g., “gpt-4o”, “gpt-4o-mini”)
   • The selected model will be displayed on the button
   

2. Model Settings:

   Click the Model Settings icon (⚙️) next to the model selector to fine-tune response generation. Settings vary by model type:

   Reasoning Models (e.g. GPT-5.4)

   Parameter	Description
   Reasoning	Controls depth of logical thinking and problem-solving.

   Level	Behavior
   Low	Fast, surface-level reasoning with concise answers and minimal steps
   Medium	Balanced reasoning with clear explanations and moderate multi-step thinking (default)
   High	Deep, thorough reasoning with detailed step-by-step analysis (may be slower)

   

   Standard Models (e.g. GPT-4o)

   Parameter	Description
   Creativity	Controls response randomness. Lower values produce focused, deterministic outputs; higher values produce more varied and creative responses.

   Level	Behavior
   1	Highly focused and deterministic outputs
   2	Mostly focused with slight variation
   3	Balanced between focus and creativity (default)
   4	More varied and creative responses
   5	Maximum creativity and diversity

   

   Max Completion Tokens (All Models)

   Option	Description
   Auto	System sets the token limit to 4096 tokens (default)
   Custom	Manually enter a specific token limit. The interface shows remaining tokens available. An error is shown if the value exceeds the model’s maximum output tokens.

   Capabilities (shown when supported by the selected model)

   Badge	Meaning
   	The model accepts image inputs alongside text
   	The model uses extended chain-of-thought reasoning

   Capability badges appear automatically at the bottom of the settings panel based on the selected model. If neither capability is supported, the Capabilities row is hidden.

---

Toolkits Configuration

In the TOOLKITS section, you can enhance your pipeline’s capabilities by adding various integrations:

The + Toolkit, + MCP, + Agent, and + Pipeline buttons are disabled until the pipeline has been saved at least once. Save the pipeline in the Configuration tab first, then add integrations.

Toolkits

• Click + Toolkit to select from available toolkits or create new ones

• Browse and select toolkits like GitHub, Jira, Confluence, etc.

  

MCP Connections

• Click + MCP to add Model Context Protocol connections

• Select from available MCP configurations or create new ones

• MCP connections appear in the toolkits list with proper status indicators

• For remote MCP servers, an Authorize button appears — click it to complete the OAuth authorization flow before the connection becomes active

  

Nested Agents

• Click + Agent to add agents as pipeline components

• Select from existing agents in your project

• Choose specific versions from the version dropdown

• Nested agents provide AI capabilities within pipeline nodes

  

Nested Pipelines

• Click + Pipeline to integrate other pipeline workflows

• Select from available pipelines in your project

• Nested pipelines enable complex workflow compositions

  

Flow Editor Tab

The Flow Editor tab provides a visual interface for designing your pipeline workflow.

The Flow Editor is disabled during pipeline creation. You must save the pipeline first in the Configuration tab before accessing the Flow Editor.

Visual Node Management

• Drag and drop nodes onto the canvas
• Connect nodes to create workflow paths
• Configure individual node properties
• Click the + (Add Node) button to add nodes to your workflow
• Node types include: Agent, Decision, Loop, Router, State Modifier, and more

State Management

• State: Holds data that flows through your workflow
• State Variables: Store information between nodes
• State Modifier Node: Update, transform, or clean up state variables during execution
• Configure input and output variables per node to control data flow

YAML Synchronization

• Flow Editor changes automatically update the YAML instructions
• Manual YAML edits are reflected in the visual editor
• Seamlessly switch between visual and code-based editing

Editor Modes

• Flow mode: Visual drag-and-drop interface
• YAML mode: Direct YAML code editing

Finalizing Pipeline Creation

Once you have completed configuring your pipeline:

• Click the Save button to save all your configuration changes
• Click the X button to close the canvas interface

Your newly created pipeline will appear in the PARTICIPANTS section under Pipelines and becomes immediately available for use in conversations.

---

Editing Pipelines via Canvas Interface

Accessing Pipeline Edit Mode

There are two ways to access the pipeline edit mode:

1. From PARTICIPANTS Section

   • Navigate to the Chat page where the pipeline is available.
   • In the PARTICIPANTS section, locate Pipelines.
   • Find the pipeline you want to edit.
   • Hover over the pipeline to reveal action buttons.
   • Click the pencil Edit icon that appears.

2. From Chat Interface

   • When a pipeline is active in your conversation, click the Pipeline Settings button (gear icon) that appears in the chat interface.
   • This will open the pipeline configuration canvas directly.

The pipeline configuration canvas will open with current settings pre-populated.

Modifying Pipeline Configuration

Once in edit mode, the interface displays current settings pre-populated. You can modify any of the following:

Configuration Tab

Welcome Message

Change the initial message users see when they start interacting with your pipeline.

Conversation Starters

Add, remove, or modify starter prompts (maximum 4) that guide users on how to interact with the pipeline.

Advanced Settings

Adjust the step limit and other execution parameters.

LLM Model and Settings

Switch to a different model or fine-tune model parameters such as creativity/reasoning level and max tokens.

Toolkits and Integrations

Add or remove toolkits, MCP connections, nested agents, and nested pipelines. Expand each entry to view and manage individual tools or select specific versions.

Flow Editor Tab

Visual Workflow Design

Add, remove, or rearrange nodes directly on the canvas.

Node Configuration

Update properties, inputs, and outputs for individual nodes.

Connection Management

Modify transitions and data flow between nodes.

YAML Editing

Directly edit the workflow YAML code for fine-grained control.

After making your desired changes, click the Save button to apply them. The updated configuration becomes immediately active.

Version Management

When editing pipelines from the chat interface:

1. Version Selector:

   • The current version is displayed on the version dropdown in the pipeline settings panel
   • The current version is also displayed on the pipeline card in the PARTICIPANTS sidebar
   • Click the version dropdown to view all available versions of the pipeline
   • Select different versions to view or edit

2. Version Status:

   • Draft versions: Can be edited and modified
   • Published versions: Read-only, cannot be edited

Published versions are read-only and cannot be edited. To modify a published version, create a new version or edit an existing draft version.

---

Troubleshooting

Save button remains disabled

Problem: Required fields are missing or incomplete.

• Review all fields marked with an asterisk (*) and ensure they are filled in
• Verify that both Name and Description fields contain valid text

Pipeline creation fails due to invalid YAML

Problem: YAML syntax error in the workflow definition.

• Verify YAML syntax — check proper indentation and structure
• Use a YAML validator or start from a simple workflow template

Pipeline creation fails due to model configuration issues

Problem: Invalid or misconfigured LLM model settings.

• Verify that a valid LLM model is selected
• Ensure model settings are within acceptable ranges
• Use default model settings initially, then customize as needed

Cannot access edit mode for existing pipeline

Problem: Insufficient permissions or version is published.

• Verify you have the applications.update permission
• Confirm the pipeline version is not in Published state (published versions are read-only)
• Ensure no active operations are currently using the pipeline

Flow Editor tab is disabled

Problem: Pipeline has not been saved yet.

• Save the pipeline first in the Configuration tab
• The Flow Editor becomes available only after the initial pipeline save

Changes cannot be saved due to validation errors

Problem: One or more fields contain invalid values.

• Review error messages displayed in the interface
• Ensure all required fields maintain valid values
• Check that external resources (models, toolkits, agents) are still accessible

Created pipeline doesn't appear in PARTICIPANTS

Problem: Pipeline may have been created in a different project context.

• Refresh the interface
• Verify the pipeline was created in the correct workspace/project
• Ensure you are viewing the correct project context

Model configuration changes don't take effect

Problem: Model settings were closed without applying.

• Reopen the model settings dialog
• Make your changes and click Apply before closing

Cannot switch between pipeline versions

Problem: Unsaved changes are blocking the version switch.

• Save or discard your current changes before switching versions
• A confirmation dialog will appear if you try to switch with unsaved changes

---

For additional information and related functionality, refer to these helpful resources:

• Pipeline Menu - Complete reference for pipeline management and configuration options
• Chat Menu - Comprehensive guide to chat interface features and navigation
• Credential Menu - Detailed instructions for managing authentication credentials
• AI Configuration - Setup and configuration guide for AI models and settings
• How to Create and Edit Agents from Canvas - Similar guide for agent management