Proof of Concept (POC): The Context Bridge
1. POC Scope
The objective of the POC is to demonstrate that an AI can identify when code submitted via a Pull Request (PR) does not meet the acceptance criteria defined in a Jira task.
2. Test Scenario
- Input A: A Jira ticket (e.g. “Create login endpoint with 2FA validation”).
- Input B: A code diff (e.g. code that creates login but omits 2FA).
- Expected output: An AI report stating: “Quality risk: the code implements login, but the 2FA logic mentioned in acceptance criterion 2 is missing.”
3. Simplified Architecture
- Mock Data: JSONs simulating Jira and GitHub APIs.
- LLM Prompting: Use Few-Shot Prompting to teach the AI how to compare requirements against code.
- Interface: A simple dashboard (Streamlit or static Markdown) comparing “Expected” vs. “Delivered”.
4. Success Metrics for the POC
- Accuracy in identifying “Requirement Gaps” above 80%.
- Analysis time under 30 seconds per ticket/PR.
Next Steps
Languages