Skip to main content

Prompt Configuration

Declarative prompt configuration using prompt.yaml for consistent LLM enhancement patterns.

prompt.yaml Structure​

operation_description:
  system_prompt: |
    You are an expert at writing concise, OpenAI-compatible tool descriptions.
    Follow these strict rules:
    1. Keep descriptions under {max_length} characters
    2. Start with what the operation does (verb + object)
    3. Add one use case: 'Use when: <scenario>'
    4. NO markdown formatting or special characters
    5. Focus on agentic purpose

  user_prompt_template: |
    API: {method} {path}
    Operation ID: {operation_id}
    Summary: {summary}
    Original Description: {description}

    Write a concise, agent-focused tool description.

parameter_description:
  system_prompt: |
    Write brief, practical parameter descriptions for AI agents.

  user_prompt_template: |
    Parameter: {param_name}
    Type: {param_type}
    Required: {param_required}
    Original: {original_description}

    Write usage guidance for this parameter.

Benefits​

  • Version Control: Prompts tracked in git and reviewable in PRs
  • Team Collaboration: Non-developers can improve prompt quality
  • A/B Testing: Easy to test different prompt strategies
  • Consistency: Same prompts used across all API operations

Documentation in progress. See examples for working prompt configurations.