Templates
Templates are pre-configured session configurations that streamline common automation workflows.
What are Templates?
Templates define:
- System Prompt: Instructions for the AI model
- Initial Message: Optional starting message
- Required Tools: Tools needed for the workflow
- Default Model: Recommended model for the task
- Variables: Placeholders for customization
Benefits of Templates
Consistency
- Standardized workflows across team
- Reproducible results
- Best practices encoded in prompts
Efficiency
- Quick start for common tasks
- No manual configuration needed
- Pre-tested configurations
Sharing
- Share effective prompts with team
- Organization-wide templates
- Public templates for community
Using Templates
Selecting a Template
- Go to Automation dashboard
- Click the template dropdown
- Browse or search templates
- Click a template to select it
- Review with Preview button
Template Preview
Preview shows:
- Template name and description
- Category and slug
- System prompt content
- Initial message (if any)
- Required tools list
- Default model
Executing a Template
- Select a gateway
- Select a template
- Choose a model (or use default)
- Configure tools (optional)
- Click Start Session
Via API
POST /api/automation/templates/:id/execute
Content-Type: application/json
Authorization: Bearer <your-api-key>
{
"gatewayIds": ["gateway-uuid-1", "gateway-uuid-2"],
"variables": {
"project_name": "my-app",
"target_language": "python"
},
"modelId": "groq/llama-3.3-70b-versatile",
"disabledTools": ["web_search"]
}Template Structure
Core Fields
| Field | Required | Description |
|---|---|---|
name |
Yes | Display name for the template |
slug |
Yes | URL-friendly identifier |
category |
Yes | Category for organization |
system_prompt |
Yes | AI system instructions |
description |
No | Brief explanation |
icon |
No | Emoji or icon |
initial_message |
No | First message to send |
required_tools |
No | Array of required tool names |
default_model |
No | Recommended model ID |
variables |
No | Variable definitions |
is_public |
No | Whether template is publicly visible |
is_active |
No | Whether template is enabled |
System Prompt
The system prompt instructs the AI model:
You are an expert code reviewer. Your task is to:
1. Analyze code for potential bugs and issues
2. Suggest performance improvements
3. Identify security vulnerabilities
4. Recommend best practices
Be thorough but constructive. Explain your reasoning clearly.Variables
Templates support variable placeholders:
You are working on the {{project_name}} project.
The primary programming language is {{language}}.
Focus on {{focus_area}}.Variables are replaced when executing the template.
Creating Templates
Via Web Interface
- Navigate to Automation Templates page
- Click Create Template
- Fill in template details:
- Name and slug
- Category
- System prompt
- Optional fields
- Save the template
Via API
POST /api/automation/templates
Content-Type: application/json
Authorization: Bearer <your-api-key>
{
"name": "Code Review Assistant",
"slug": "code-review",
"category": "development",
"icon": "🔍",
"description": "Automated code review and suggestions",
"system_prompt": "You are an expert code reviewer...",
"initial_message": "Please share the code you'd like me to review.",
"required_tools": ["read_file", "list_files", "search_files"],
"default_model": "groq/llama-3.3-70b-versatile",
"variables": {
"language": {
"type": "string",
"default": "python",
"description": "Primary programming language"
}
},
"is_public": false,
"is_active": true
}Managing Templates
List Templates
GET /api/automation/templates?category=development&limit=50
Authorization: Bearer <your-api-key>Get Template Details
GET /api/automation/templates/:id
Authorization: Bearer <your-api-key>Update Template
PUT /api/automation/templates/:id
Content-Type: application/json
Authorization: Bearer <your-api-key>
{
"name": "Updated Name",
"system_prompt": "Updated instructions...",
"default_model": "anthropic/claude-3-5-sonnet"
}Delete Template
DELETE /api/automation/templates/:id
Authorization: Bearer <your-api-key>Template Categories
Development
- Code review
- Bug analysis
- Refactoring assistance
- Documentation generation
DevOps
- Deployment scripts
- Infrastructure checks
- Log analysis
- Health monitoring
Data
- Data transformation
- Report generation
- Analysis workflows
- Batch processing
Research
- Information synthesis
- Summary generation
- Comparative analysis
- Trend identification
Template Variables
Defining Variables
{
"variables": {
"target_directory": {
"type": "string",
"required": true,
"description": "Directory to analyze"
},
"max_files": {
"type": "number",
"default": 100,
"description": "Maximum files to process"
},
"include_tests": {
"type": "boolean",
"default": false,
"description": "Include test files"
}
}
}Variable Types
| Type | Description |
|---|---|
string |
Text value |
number |
Numeric value |
boolean |
True/false |
array |
List of values |
Using Variables in Prompts
Analyze the code in {{target_directory}}.
Process up to {{max_files}} files.
{{#if include_tests}}Include test files in the analysis.{{/if}}Providing Variable Values
POST /api/automation/templates/:id/execute
Content-Type: application/json
{
"gatewayIds": ["gateway-uuid"],
"variables": {
"target_directory": "/app/src",
"max_files": 50,
"include_tests": true
}
}Required Tools
Specifying Tools
List tools the template needs:
{
"required_tools": [
"read_file",
"list_files",
"search_files",
"write_file"
]
}Tool Validation
When executing:
- Gateway must support all required tools
- Warning shown if tools are missing
- Session may still proceed
Common Tool Sets
| Workflow | Recommended Tools |
|---|---|
| Code Analysis | read_file, list_files, search_files |
| Code Modification | Above + write_file, create_directory |
| Web Research | web_search, web_browse, read_url |
| System Admin | execute_command, read_file, write_file |
Best Practices
Writing System Prompts
- Be specific: Clearly define the AI's role and capabilities
- Set expectations: Describe desired output format
- Add constraints: Specify what not to do
- Include examples: Show expected responses
Example System Prompt
You are an expert Python developer specializing in code optimization.
Your responsibilities:
- Analyze Python code for performance issues
- Suggest specific optimizations with code examples
- Explain the impact of each optimization
- Consider memory usage and execution time
Output format:
1. Summary of findings
2. Prioritized list of optimizations
3. Code examples for each suggestion
Constraints:
- Only suggest changes that maintain functionality
- Do not modify external dependencies
- Preserve existing code styleTemplate Organization
- Use consistent naming conventions
- Group by category
- Include descriptive slugs
- Add helpful descriptions
Testing Templates
- Create template with test prompt
- Execute on test gateway
- Review results
- Iterate on system prompt
- Publish when satisfied
Maintenance
- Review templates periodically
- Update for new model capabilities
- Remove obsolete templates
- Gather user feedback
Sharing Templates
Organization Templates
Templates can be scoped to your organization:
- Visible to all organization members
- Managed by admins
- Share team best practices
Public Templates
Make templates public for community use:
- Set
is_public: true - Add clear documentation
- Include usage examples
- Maintain actively
Template Examples
Code Review Template
{
"name": "Code Review",
"slug": "code-review",
"category": "development",
"icon": "🔍",
"system_prompt": "You are a senior software engineer conducting a code review. Analyze the code for:\n\n1. **Bugs and Logic Errors**: Identify potential issues\n2. **Performance**: Suggest optimizations\n3. **Security**: Flag vulnerabilities\n4. **Best Practices**: Recommend improvements\n5. **Readability**: Suggest clarity improvements\n\nProvide specific line references and code examples.",
"required_tools": ["read_file", "list_files", "search_files"],
"default_model": "anthropic/claude-3-5-sonnet"
}Documentation Generator
{
"name": "Documentation Generator",
"slug": "doc-gen",
"category": "development",
"icon": "📝",
"system_prompt": "You are a technical writer. Generate comprehensive documentation for the provided code:\n\n1. Overview and purpose\n2. Installation instructions\n3. API reference\n4. Usage examples\n5. Configuration options\n\nUse Markdown format with clear headings.",
"initial_message": "I'll analyze the codebase and generate documentation. Please specify the project directory.",
"required_tools": ["read_file", "list_files", "write_file"],
"default_model": "groq/llama-3.3-70b-versatile"
}System Health Check
{
"name": "System Health Check",
"slug": "health-check",
"category": "devops",
"icon": "🏥",
"system_prompt": "You are a DevOps engineer performing a system health check. Check:\n\n1. Disk usage and available space\n2. Memory utilization\n3. Running processes\n4. Network connectivity\n5. Service status\n\nReport issues with severity levels (critical/warning/info).",
"required_tools": ["execute_command", "read_file"],
"default_model": "groq/llama-3.3-70b-versatile"
}Next Steps
- Inference Sessions - Managing automation sessions
- Automation Overview - Back to automation overview
- Models Guide - Choosing the right model