dhirajb05/opensprint

Purpose and Entry Points

GitHub Roast Machine is a React/Vite single-page application backed by Express serverless/API handlers. A user submits a GitHub username, receives an AI-generated roast, career score, developer archetype, meme content, and a downloadable/shareable card. Frontend bootstraps at Github-Roast/Profile_Analyser/src/main.jsx; routing and app composition begin in Github-Roast/Profile_Analyser/src/App.jsx. The local backend entry point is Github-Roast/Profile_Analyser/server/index.js; Vercel integration is represented by Github-Roast/Profile_Analyser/api/index.js and Github-Roast/Profile_Analyser/vercel.json.

Frontend Interaction Flow

The landing screen collects the target username and roast mode in Github-Roast/Profile_Analyser/src/pages/Landing.jsx. The app transitions through terminal-style progress UI in Github-Roast/Profile_Analyser/src/pages/Loading.jsx, then presents generated content in Github-Roast/Profile_Analyser/src/pages/Results.jsx. Client navigation is React Router-based, while shared retro visual styling is concentrated in Github-Roast/Profile_Analyser/src/index.css and Github-Roast/Profile_Analyser/src/App.css.

GitHub Profile Acquisition Boundary

The backend’s GitHub integration is isolated in Github-Roast/Profile_Analyser/server/routes/github.js. This route fetches and shapes public GitHub profile/repository activity into input suitable for analysis. It forms the external-data boundary: frontend requests backend data rather than directly relying on GitHub from browser code, allowing credentials, normalization, rate-limit behavior, and API failures to remain server-side.

AI Roast Generation and Caching

Github-Roast/Profile_Analyser/server/routes/roast.js owns the main analysis stage: it turns GitHub-derived facts plus the chosen tone (Friendly, Sarcastic, or Savage) into the roast, score, and archetype through Groq’s Llama 3.1 8B Instant API. This route also implements the stated 24-hour caching behavior, using in-memory cache by default and optionally Vercel KV/Redis in production. Required service credentials and deployment-time configuration are indicated by Github-Roast/Profile_Analyser/server/.env.example.

Meme Generation Stage

Github-Roast/Profile_Analyser/server/routes/meme.js is a separate generation boundary for selecting meme templates and producing roast-specific captions. Keeping memes outside the primary roast route allows the results experience to compose optional/generated visual content independently of profile acquisition and core roast analysis. Rendered meme editing is handled by Github-Roast/Profile_Analyser/src/pages/MemeEditor.jsx.

Results, Export, and Sharing

Github-Roast/Profile_Analyser/src/pages/Results.jsx composes roast output, score, archetype, and inline meme results for presentation and social sharing. Github-Roast/Profile_Analyser/src/pages/RoastCard.jsx provides the styled, downloadable roast-card representation. These are presentation/export boundaries: generated backend payloads become user-controlled visual artifacts without moving AI or GitHub API access into the client.

Runtime and Deployment Topology

The frontend runs as a Vite-built React 19 application (Github-Roast/Profile_Analyser/package.json). The backend runs on Node.js with Express (Github-Roast/Profile_Analyser/server/package.json) during local development, while Vercel routes deployment through Github-Roast/Profile_Analyser/api/index.js. Material external services are GitHub’s API for source profile data, Groq for LLM generation, and optional Vercel KV Redis for persistent production caching.