prompts.yaml¶
The config/prompts.yaml
file contains all AI prompts and templates used by Dev Agents. Customize these to change how your agent behaves and responds.
File Structure¶
Prompts are organized by agent type and function:
# System prompts for different agent types
agents:
git_chatbot:
system_prompt: |
You are BettySharp, a senior software engineer...
code_research:
system_prompt: |
You are an expert code analyst...
impact_analysis:
system_prompt: |
You specialize in analyzing code changes...
# Templates for specific responses
templates:
code_review:
summary: |
## Code Review Summary for {pr_title}
impact_report:
header: |
# Impact Analysis Report
**Changes analyzed:** {changed_files_count} files
Agent System Prompts¶
Git Chatbot Agent¶
agents:
git_chatbot:
system_prompt: |
You are BettySharp, a senior software engineer and helpful assistant.
Your role is to help developers understand codebases, analyze changes,
and provide insights about software development practices.
Key capabilities:
- Analyze git diffs and code changes
- Explain code functionality and patterns
- Suggest improvements and best practices
- Generate testing recommendations
Always provide practical, actionable advice.
Code Research Agent¶
agents:
code_research:
system_prompt: |
You are an expert code analyst with deep knowledge of software architecture.
Your task is to research and understand codebases by:
- Reading and analyzing source code files
- Understanding code relationships and dependencies
- Identifying patterns and architectural decisions
- Explaining complex code logic clearly
Provide thorough but concise explanations.
Impact Analysis Agent¶
agents:
impact_analysis:
system_prompt: |
You specialize in analyzing the impact of code changes.
For each change, assess:
- Functional impact: What behavior changes
- Technical impact: Architecture, performance, security implications
- Testing requirements: What needs to be tested
- Risk assessment: Potential breaking changes or issues
Provide structured, actionable impact reports.
Response Templates¶
Code Review Template¶
templates:
code_review:
summary: |
## Code Review Summary for {pr_title}
**Files Changed:** {changed_files_count}
**Lines Added:** {lines_added} | **Lines Removed:** {lines_removed}
### Overview
{overview}
### Key Changes
{key_changes}
### Recommendations
{recommendations}
### Testing Suggestions
{testing_suggestions}
Impact Analysis Template¶
templates:
impact_analysis:
header: |
# Impact Analysis Report
**Generated:** {timestamp}
**Changes analyzed:** {changed_files_count} files
ui_impact: |
## 🎨 UI/UX Impact
**Risk Level:** {risk_level}
{description}
api_impact: |
## 🔌 API Impact
**Breaking Changes:** {has_breaking_changes}
{description}
footer: |
---
*Generated by Dev Agents - Impact Analysis*
Slack Message Templates¶
slack:
messages:
analysis_complete: |
✅ **Analysis Complete**
I've analyzed the changes in {context}
{summary}
Would you like me to elaborate on any aspect?
error_response: |
❌ **Error Processing Request**
I encountered an issue: {error_message}
Please check the configuration and try again.
Customization¶
Changing Agent Personality¶
Modify the system prompts to change how your agent behaves:
agents:
git_chatbot:
system_prompt: |
You are DevBot, a friendly and enthusiastic coding assistant.
You love helping developers and always provide encouragement
along with technical advice. Use emojis and keep responses upbeat! 🚀
Adding New Templates¶
Add custom templates for new use cases:
templates:
deployment_checklist:
header: |
## 🚀 Deployment Checklist for {release_version}
items: |
- [ ] Database migrations applied
- [ ] Environment variables updated
- [ ] Tests passing
- [ ] Documentation updated
Environment Variable Substitution¶
Prompts support environment variable substitution:
agents:
git_chatbot:
system_prompt: |
You are ${AGENT_NAME:-BettySharp}, working with the ${AGENT_TEAM:-development} team.
Validation¶
Test your prompt changes:
python -c "
from src.core.prompts import BasePrompts
prompts = BasePrompts()
print('✓ Prompts loaded successfully')
print('Git chatbot prompt:', prompts.get_value('agents.git_chatbot.system_prompt')[:100])
"
Best Practices¶
- Keep prompts focused - Each agent should have a clear, specific role
- Use consistent formatting - Maintain template structure across agents
- Include context - Provide enough background for accurate responses
- Test changes - Verify prompt modifications work as expected
- Version control - Track prompt changes like any other code
Next Steps¶
- Configure integrations
- Test your customizations with the Quick Start guide
- Review environment variables for additional customization