Weekend Showcase Challenge: BugPilot AI — AI-Powered QA Bug Analyzer & Test Generator
BugPilot AI is an intelligent QA copilot powered by Amazon Bedrock and AWS Lambda that transforms raw stack traces into structured triage, reproduction steps, test case suites, and executable Playwright automation scripts in second
1. Vision & What It Does
In modern software development, bridging the gap between raw, unstructured user bug reports and actionable engineering tasks is often a manual, time-consuming bottleneck. Developers and QA engineers spend valuable hours reading cryptic error messages, asking users for reproduction steps, triaging severity, writing manual test cases, and building regression automation scripts.
BugPilot AI solves this problem by acting as an intelligent AI QA Copilot.
When a tester or developer pastes a bug description, console error trace, or testing observation into BugPilot AI, the application leverages Amazon Bedrock to instantly generate a comprehensive, structured QA specification:
- Instant Severity & Priority Triage: Automatically assigns clear severity levels (Critical, High, Medium, Low) and priority rankings (P0 through P3).
- Category Classification: Classifies issues into core architectural domains like UI/UX State, Performance, Authentication & Security, or AWS Infrastructure.
- Numbered Reproduction Steps: Turns ambiguous bug descriptions into clear, step-by-step reproduction instructions.
- Speculative Root Cause Analysis: Generates hypothesis-backed potential root causes to help developers pinpoint the bug faster.
- Positive & Negative Test Cases: Produces structured test specifications with preconditions, execution steps, expected outcomes, and priority labels.
- Interactive Regression Checklist: Generates a real-time interactive checklist to verify that surrounding features remain stable post-fix.
- Executable Playwright Test Generator: Automatically scaffold clean, syntax-highlighted Playwright TypeScript/JavaScript end-to-end automation scripts with one-click clipboard copying.
By automating the heavy lifting of QA triage and automation scaffold creation, BugPilot AI reduces bug triage cycles from hours to seconds.
2. How You Built It
The development process for BugPilot AI was guided by a single principle: maximum developer productivity with zero unnecessary friction.
Key Architectural & Design Decisions:
- Developer-First UI/UX: Built with React, Vite, and custom CSS design tokens. We designed a dark glassmorphism interface with high-contrast badge indicators, glowing status indicators, and keyboard-friendly controls rather than a basic chat interface.
- Serverless & Cost Efficiency: Selected AWS Lambda combined with Amazon Bedrock to ensure zero baseline infrastructure cost when idle, remaining 100% within the AWS Free Tier.
- Structured Schema Enforcement: Configured Amazon Bedrock with system instructions enforcing strict JSON output schemas. This eliminates malformed responses and allows the React frontend to render rich, interactive UI components directly from AI outputs.
- Instant Showcase Resilience: Built a dual-mode engine with pre-loaded realistic scenarios (Checkout Blank Screen, OAuth Safari Redirect Loop, 4K Image Upload Memory Crash). This guarantees that reviewers can experience the full power of BugPilot AI instantly even without local AWS API credentials.
Challenges Encountered & Solutions:
- Challenge (JSON Output Parsing): LLMs sometimes prepend or append markdown code blocks (e.g. ```json) around JSON output, which causes standard
JSON.parsecalls to fail.
- Solution: Implemented robust regex sanitation inside the AWS Lambda handler (
index.mjs) to strip code block fences and validate JSON schema before returning HTTP responses.
3. AWS Services Used & Architecture Overview
BugPilot AI relies on a clean, scalable serverless AWS architecture:
[ Frontend: React + Vite ]
│
▼ (HTTP POST /analyze)
[ Amazon API Gateway ]
│
▼ (Lambda Event Invocation)
[ AWS Lambda (Node.js 20.x) ]
│
▼ (InvokeModel Command)
[ Amazon Bedrock (Amazon Nova / Bedrock Models) ]AWS Services Breakdown:
- Amazon Bedrock: Serves as the core generative AI foundation model provider. It performs deep reasoning over raw stack traces and outputs structured QA JSON objects.
- AWS Lambda: Serverless execution backend executing the Node.js 20.x handler. Handles API requests, constructs Bedrock payloads, and enforces CORS and error boundaries.
- Amazon API Gateway: Provides the secure HTTP API endpoint (
https://wpmt3yl1l8.execute-api.us-east-1.amazonaws.com) connecting the client application to the AWS Lambda backend.
- AWS Amplify: Hosts the React Vite single-page application with continuous web hosting at
https://main.d359ragcc4ietx.amplifyapp.com.
- AWS Free Tier Integration: Fully compatible with AWS Free Tier credit grants, utilizing low-latency serverless resources with zero idle cost.
4. What You Learned
Participating in the Summer Build Series and the Showcase Weekend Challenge provided invaluable hands-on insights:
- Generative AI for QA Engineering: Discovered how specialized system prompts convert unstructured LLM outputs into strictly typed, reliable JSON schema structures that drive complex frontend state.
- Amazon Bedrock Integration: Learned how to seamlessly invoke Amazon Bedrock models (
client-bedrock-runtime) within AWS Lambda using AWS SDK v3.
- Serverless Resiliency: Gained experience designing serverless applications that handle timeouts, fallback modes, and graceful degradation during network disruptions.
5. Link to App & Public Repository
- Public GitHub Repository: https://github.com/anky8848/bugpilot-ai
- Live Deployed Application: https://main.d359ragcc4ietx.amplifyapp.com
Enjoyed reading this content? Let the author know!
Your likes, comments, shares, and saves help creators reach more builders.
Loading recommendations
Loading article