aanishnithin07/gitdna

Purpose and Entry Points

GitDNA is a GitHub behavioral-intelligence product: it fetches public GitHub telemetry, derives developer activity signals, and uses LLMs to produce profile narratives, traits, comparisons, and visual artifacts. The React client starts at src/main.jsx and composes the experience in src/App.jsx. The Python API is exposed through backend/main.py; api/index.py likely provides the Vercel-facing API entry.

Frontend Application Shell

src/App.jsx owns the scan/comparison workflow, profile state, and composition of the feature views. src/utils/backendApi.js is the frontend boundary for calling the backend rather than embedding API logic across UI components. Styling and application-level visual behavior live in src/App.css and src/index.css.

Analysis and AI Pipeline

backend/main.py is the FastAPI HTTP boundary: it accepts requested GitHub identities/features and returns analysis results. backend/analyzer.py is the deterministic analysis stage that transforms GitHub REST/Event data into behavioral metrics such as activity, timing, language, and commit-derived signals. backend/ai_engine.py is the generative interpretation layer, converting computed telemetry into archetypes, roast/battle narratives, time-machine stories, and related AI content.

GitHub and LLM Service Boundaries

The backend integrates GitHub REST and Events APIs as the source of public profile, repository, commit, and event data. LLM generation depends on Groq/Llama 3.1 and Google Gemini Flash, with credentials/configuration expected via backend/.env.example and .env.example. The key boundary is: external GitHub data is normalized and scored before being passed to AI generation, keeping visual clients insulated from provider-specific calls.

Feature Presentation Modules

Specialized React components render distinct projections of the common profile-analysis result: repository/trait analysis in src/components/RepoAnalysis.jsx, activity geography in src/components/GitMap.jsx, longitudinal narratives in src/components/TimeMachine.jsx, and workload risk in src/components/BurnoutDetail.jsx. Comparison and collectible outputs are isolated in src/components/DeveloperMatcher.jsx and src/components/TradingCard.jsx; editorial/linguistic presentations are in src/components/GitNewspaper.jsx and src/components/LanguagePrediction.jsx.

Primary Control Flow

A user supplies one or two GitHub usernames in the client. src/App.jsx invokes the backend through src/utils/backendApi.js; FastAPI coordinates GitHub retrieval, metric analysis, and optional AI enrichment. The resulting structured profile is retained client-side and fanned out to visualization components, avoiding separate external fetches per card. Share/deep-link behavior is primarily a frontend concern, while authoritative data acquisition and model calls remain server-side.

Runtime and Deployment

The frontend is a Vite/React static application, configured by vite.config.js and deployed through vercel.json. The backend is a Python/FastAPI service with dependencies in backend/requirements.txt, runtime declaration in backend/runtime.txt, and Railway deployment metadata in backend/railway.json and backend/Procfile. This creates a split deployment: Vercel serves the interactive client, while Railway hosts long-running API and AI-provider integration.