PrismGit

Transform Git commit history into coherent narrative stories. A full-stack analytics dashboard that tells the visual story of a Git repository.

Stack

• Backend: Java 17+, Spring Boot 3.4.1, Spring Data JPA, Flyway, JGit 7.1
• Frontend: Angular 19 standalone, TypeScript, RxJS
• Database: H2 (dev), PostgreSQL (prod)
• Infra: Docker, Docker Compose, Railway
• Docs: Swagger/OpenAPI (/swagger-ui.html)
• Testing: JUnit 5, Mockito (backend), Jasmine/Karma (frontend)

Features

• Repo management — Register Git repositories (local path or remote URL)
• Commit sync — Import commits via JGit (local) or GitHub REST API (remote, paginated)
• Analytics dashboard — Summary stats, timeline bars, author bars, file heatmap, hourly/daily activity heatmaps
• Story generation — Two modes:
  • Template — Groups commits by date and author into markdown narratives
  • LLM — AI-powered stories via OpenAI (configurable model, graceful fallback)
• REST API — 14 endpoints, fully documented via Swagger
• SPA routing — Built-in Angular app with resource handler fallback
• Logging — SLF4J/@Slf4j with contextual levels (INFO, DEBUG, WARN, ERROR)

Structure

prismgit/
├── backend/                  # REST API + business logic
│   └── src/main/java/com/thiz/prismgit/
│       ├── config/           # Spring configuration (SPA, OpenAPI, app)
│       ├── controller/       # REST controllers (Repo, Analytics, Story)
│       ├── dto/              # Data Transfer Objects
│       ├── entity/           # JPA entities (GitRepo, CommitEntry, Story)
│       ├── exception/        # Global exception handler
│       ├── repository/       # Spring Data repositories
│       └── service/          # Business logic + generators
│           └── generator/    # Template & LLM story generators
├── frontend/                 # Angular 19 SPA
│   └── src/app/
│       ├── models/           # TypeScript interfaces
│       ├── pages/            # Repos, Dashboard, StoryView
│       └── services/         # HTTP service
└── docker/                   # Docker Compose dev setup

API Endpoints

Method	Path	Description
POST	/api/repos	Register a new repository
GET	/api/repos	List all repositories
GET	/api/repos/{id}	Get repository details
DELETE	/api/repos/{id}	Delete a repository
POST	/api/repos/{id}/sync	Import commits
GET	/api/repos/{id}/commits	List commits (paginated)
GET	/api/repos/{id}/analytics/summary	Analytics summary
GET	/api/repos/{id}/analytics/timeline	Daily commit timeline
GET	/api/repos/{id}/analytics/activity/hour	Hourly distribution
GET	/api/repos/{id}/analytics/activity/day	Day-of-week distribution
POST	/api/repos/{repoId}/stories	Generate a story
GET	/api/repos/{repoId}/stories	List stories for a repo
GET	/api/stories	List all stories
GET	/api/stories/{id}	Get a story

Full interactive docs at /swagger-ui.html when running.

Configuration

Environment variables

Variable	Description	Required
GITHUB_TOKEN	GitHub personal access token	For remote sync
LLM_API_KEY	OpenAI API key	For LLM story mode
LLM_MODEL	OpenAI model (default: gpt-4o-mini)	No
LLM_PROVIDER	LLM provider (default: openai)	No
PGHOST, PGPORT, PGDATABASE, PGUSER, PGPASSWORD	PostgreSQL connection (prod)	Production

Profiles

• Default (dev): H2 in-memory database, Flyway migrations
• prod: PostgreSQL, Flyway clean disabled

Development

Prerequisites

• Java 17+, Maven
• Node.js 22+, npm

Run backend (dev mode)

cd backend
./mvnw spring-boot:run

API at http://localhost:8080 with H2 in-memory DB. H2 console: /h2-console.

Run frontend (dev mode)

cd frontend
npm install
ng serve

Frontend at http://localhost:4200, proxies API calls to localhost:8080.

Run with Docker Compose (full stack)

docker compose -f docker/docker-compose.yml up

Testing

Backend (48+ tests): JUnit 5 + Mockito

• Controllers: Repo, Analytics, Story (20 tests)
• Services: Repo, Story, LLM Generator, RemoteSync (28+ tests)
• Mocked OpenAI and GitHub APIs

Frontend (52+ tests): Jasmine + Karma

• Components: Dashboard, Repos, StoryView (32 tests)
• Service: ApiService (20 tests)

# Backend
cd backend && mvn test

# Frontend
cd frontend && ng test

Deploy

Railway

The project auto-deploys from the main branch via Dockerfile (multi-stage: Angular build → Maven build → JRE runtime).

railway up --service prismgit

License

MIT