Skip to main content

​
Importing & Exporting AI Connection Configurations

Share AI connection configurations across organizations or create backups with YAML files.

​
Overview

Export your AI connection configurations as YAML files to:
  • Share AI connection setups with team members or other organizations
  • Backup important configurations
  • Document your AI connection architecture
  • Template common configurations for reuse
Security Note: Exported configurations never include sensitive data. API keys and header values are excluded for security.

​
Exporting an AI Connection Configuration

​
Step-by-Step Export

  1. Navigate to Admin Dashboard → AI Connections tab
  2. Find the AI connection you want to export
  3. Hover over the connection card
  4. Click the Download icon (↓)
  5. YAML file downloads automatically
Filename format: {connection-name}-config-{date}.yaml

​
What Gets Exported

✅ Included:
  • AI connection name and description
  • API endpoint URL
  • HTTP method (POST, GET, etc.)
  • Header keys (without values)
  • Request schema/body structure
  • Response path configuration
  • Message formatting rules
  • Suggestion prompts
❌ Never Exported (Security):
  • API keys
  • Authentication tokens
  • Header values (only keys are exported)
  • Credentials of any kind

​
Example Export

name: "Claude Sonnet"
model_id: "claude-3-sonnet-20240229"
description: "Large model for reasoning tasks"
endpoint: "https://api.anthropic.com/v1/messages"
method: "POST"

headers:
  x-api-key:          # Value not included
  anthropic-version:  # Value not included
  Content-Type:       # Value not included

request_schema:
  model: "claude-3-sonnet-20240229"
  messages:
    - role: "user"
      content: "{{message}}"
  max_tokens: 4096

response_path: "content[0].text"

message_format:
  preset: "openai"
  mapping:
    role:
      source: "role"
      target: "role"
      transform: "none"
    content:
      source: "content"
      target: "content"
      transform: "none"

suggestion_prompts:
  - "Tell me about your capabilities"
  - "How can you help me today?"
Notice that header values are empty - only the keys are included to show what headers are required.

​
Importing an AI Connection Configuration

​
Step-by-Step Import

  1. Navigate to Admin Dashboard → AI Connections tab
  2. Click the Import button (top-right)
  3. Upload your YAML file:
    • Drag and drop, or
    • Click to browse files
  4. Click Validate to check the configuration
  5. Review any validation errors
  6. Click Import AI Connection to create the connection

​
After Import: Required Steps

1

Add API Keys

Navigate to the imported AI connection settings and add the required API key
2

Configure Headers

Fill in header values that were excluded from the export:
  • Authentication tokens
  • API versions
  • Content types
3

Test the Connection

Send a test message to verify the configuration works correctly

​
Import Options

Enable this option to update an existing AI connection with the same name instead of creating a duplicate.
  • ✅ Useful for updating configurations
  • ⚠️ Will overwrite existing settings
This option is currently not available for security reasons.
  • API keys must be added manually after import
  • This prevents accidental credential sharing

​
Validation Errors

Common validation errors and solutions:
Error: YAML parsing failedSolution: Validate your YAML file with a YAML validator before importing
Error: Required field is missingSolution: Ensure these required fields are present:
  • name
  • endpoint
  • request_schema
  • response_path
  • message_format
Error: Invalid endpoint URLSolution: Ensure endpoint is a valid HTTPS URL
Error: AI connection with this name already existsSolution: Either:
  • Enable “Replace if exists” option, or
  • Change the AI connection name in your YAML file

​
Sharing Configurations

​
Sharing Best Practices

1

Export the AI connection

Download the YAML configuration file
2

Review the file

Verify no sensitive data is included (it shouldn’t be, but always check)
3

Share securely

Send the YAML file via:
  • Email
  • Slack/Teams
  • Git repository
  • Documentation
4

Provide credentials separately

Send API keys and header values through a secure channel:
  • Password manager
  • Secure credential sharing service
  • Encrypted communication
Never send API keys or credentials in the same channel as the configuration file!

​
Use Cases

​
Team Collaboration

Share successful AI connection configurations with your team: 
# marketing-assistant.yaml
name: "Marketing Content Assistant"
description: "Specialized for creating marketing content"
# ... rest of configuration
Team members can import and use immediately (after adding their API keys).

​
Environment Migration

Move configurations between environments:
  1. Development → Export configuration
  2. Staging → Import and test
  3. Production → Import verified configuration

​
Configuration Templates

Create templates for common AI connection types:
  • Customer Support Bot - Pre-configured for support scenarios
  • Code Assistant - Optimized for code generation
  • Content Writer - Tuned for creative writing

​
Backup & Disaster Recovery

Regular backups of critical configurations:
  1. Export all AI connections weekly
  2. Store in version control
  3. Quick recovery if needed

​
Security Considerations

​
What’s Protected

Never Exported

  • API keys
  • Authentication tokens
  • Header values
  • Credentials
  • Secrets

Always Exported

  • AI connection name
  • Endpoint URL
  • Header keys
  • Request structure
  • Configuration settings

​
Security Best Practices

✅ Do:
  • Review exported files before sharing
  • Share credentials separately and securely
  • Use version control for configuration files
  • Document which API keys are needed
❌ Don’t:
  • Manually add API keys to YAML files
  • Share configuration and credentials together
  • Commit credentials to git repositories
  • Share files publicly without review

​
Troubleshooting

Symptoms: Download doesn’t start or failsSolutions:
  • Check browser console for errors
  • Verify you have permission to access the AI connection
  • Try a different browser
  • Ensure AI connection exists in database
Symptoms: Validation errors during importSolutions:
  • Check error messages for specific issues
  • Validate YAML syntax with online validator
  • Ensure all required fields are present
  • Verify endpoint URL is valid HTTPS
Symptoms: Imported AI connection fails to respondSolutions:
  • Verify API key was added correctly
  • Check all header values are filled in
  • Test the endpoint URL directly
  • Review request schema format
Symptoms: Don’t know what values to use for headersSolutions:
  • Check API provider documentation
  • Review the original model configuration
  • Contact person who shared the config
  • Check with API provider support

Next: Test Your Agents

Learn how to properly test imported AI connection configurations
⌘I