Skip to content

chatgpt-setup.md

Latest commit

 

History

History
92 lines (67 loc) · 3.4 KB

chatgpt-setup.md

File metadata and controls

92 lines (67 loc) · 3.4 KB
Raw

ChatGPT Integration Setup

GitFlow Analytics now supports ChatGPT-4.1 for generating qualitative executive summaries in the narrative report.

Setup

  1. Get an OpenAI API Key

  2. Set the Environment Variable

    Add to your .env file (in the same directory as your config.yaml):

    OPENAI_API_KEY=sk-...your-api-key-here...

    Or export it in your shell:

    export OPENAI_API_KEY="sk-...your-api-key-here..."

  3. Run Analysis

    When you run the analysis with markdown output enabled, ChatGPT will automatically generate a qualitative executive summary:

    gitflow-analytics analyze -c config.yaml --weeks 4

What ChatGPT Analyzes

The ChatGPT integration provides:

  • Executive Summary: 2-3 paragraphs highlighting overall team performance
  • Strategic Insights: Deep analysis of team dynamics and project health
  • Actionable Recommendations: 3-5 specific recommendations for leadership
  • Risk Assessment: Identification of potential issues and mitigation strategies

Output

The ChatGPT analysis appears in the narrative report under a new "Qualitative Analysis" section after the standard executive summary metrics.

Cost Considerations

  • Each analysis uses approximately 1,000-1,500 tokens
  • With GPT-4-turbo pricing, this costs approximately $0.01-0.02 per analysis
  • The integration only runs when OPENAI_API_KEY is set

Fallback Behavior

If ChatGPT is unavailable or fails:

  • A basic algorithmic summary is generated instead
  • The analysis continues without interruption
  • A warning message indicates the fallback was used

Example Output

## Qualitative Analysis

Over the past 4 weeks, the development team has demonstrated exceptional momentum with a 
significant increase in velocity, delivering 280 commits across 13 active projects. The 
team health score of 75/100 reflects a well-functioning unit, though there are opportunities 
for improvement in cross-team collaboration.

The concentration of work among top contributors (Luca-Borda with 30.4% of commits) suggests 
strong technical leadership but also indicates potential knowledge silos. The 34.3% ticket 
coverage reveals a gap in process adherence that could impact traceability and project 
management effectiveness.

### Strategic Insights

1. **Productivity Patterns**: The team shows consistent afternoon productivity peaks, 
   suggesting potential for morning standups or collaborative sessions
2. **Project Distribution**: FRONTEND_PROJECT dominates with 77.3% of activity, 
   indicating either strategic focus or resource imbalance
3. **Developer Specialization**: High focus scores (100% for several developers) suggest 
   deep expertise but limited knowledge sharing

### Recommendations

1. **Implement Pair Programming**: Address knowledge concentration by rotating developers 
   across projects
2. **Enhance Ticket Discipline**: Target 60%+ ticket coverage through pre-commit hooks 
   or workflow automation
3. **Balance Workload**: Consider redistributing tasks from high-volume contributors to 
   prevent burnout
4. **Morning Collaboration**: Leverage natural afternoon productivity by scheduling 
   collaborative work in mornings
5. **Cross-Project Reviews**: Institute weekly cross-project code reviews to break down silos