Docs
The web app runs repo and user analysis, persists reports, and presents a compact grading summary. The analyzer service owns the scoring logic and training pipeline.
Workspace
`apps/web` holds the Next.js dashboard and API routes.
`services/analyzer` contains the Python scoring and training code.
`datasets` stores labeled commit datasets and manifests.
`reports` contains sample and generated report artifacts.
Development
Install dependencies, then start the web app from the repo root.
npm install
npm run dev:webConnect
User analysis now starts with a GitHub App install and authorization flow.
The app redirects users to GitHub, stores the returned session server-side, and loads granted repositories.
Selected repos are sent to the analyzer so the report targets only the repositories you chose.
Analyzer
cd services/analyzer
python -m venv .venv
.venv\Scripts\activate
pip install -e .
uvicorn gitgrade_analyzer.main:app --reloadAPI
POST /api/analyze/repo
{
"repo": "vercel/next.js",
"commit_limit": 40
}API
POST /api/analyze/user
{
"username": "aaravhmodi",
"selected_repos": ["aaravhmodi/gitgrade", "aaravhmodi/news-scraper"],
"repo_limit": 2,
"commits_per_repo": 30
}Direction
Build the rule-based analyzer first.
Export stable JSON reports.
Visualize those reports in the web app.
Add labeled data and ML once the scoring contract is stable.