Layercode Product Release Reports

Full-stack Next.js app that generates internal release notes by pulling live commit data from GitHub and summarising it with the OpenAI GPT-5 model via the Vercel AI SDK.

Features

• Single endpoint (GET /deploy-report) that validates the requested date window and enforces a 30-day cap.
• Live GitHub fetch using the provided access token; merge commits are automatically filtered out.
• LLM summary generation via Vercel AI SDK + openai/gpt-5, returning semantic HTML without background jobs.
• Responsive React UI that matches the original colour palette while upgrading to the Next.js app router.
• Spec-compliant error handling for validation, rate limits, upstream failures, and the 120 s timeout budget.

Getting Started

1. Install dependencies

   npm install

2. Configure environment variables

   cp .env.local.example .env.local
   # then edit .env.local with real values

   Required keys:

   • GITHUB_TOKEN – token with read access to the target repo
   • GITHUB_REPO – OWNER/REPO
   • OPENAI_API_KEY – OpenAI API key with access to GPT-5 (Responses API)
   • OPENAI_MODEL – e.g. gpt-5 (falls back to OPENAI_REALTIME_MODEL for backwards compatibility).
     Use the bare model name—any provider prefix like openai/ is stripped automatically.

3. Run locally

   npm run dev

   The app serves both the UI and the /deploy-report route at http://localhost:3000.

API Contract

GET /deploy-report?start=YYYY-MM-DD&end=YYYY-MM-DD

• Validates input format, enforces start <= end, and rejects ranges > 30 days (413).
• Queries GitHub commits in [startT00:00:00Z, endT23:59:59Z], paginating per_page=100.
• Drops merge commits (multiple parents or messages that begin with “Merge”).
• Builds compact bullets and requests a semantic HTML report from OpenAI GPT-5 via Vercel AI SDK.

200 Response

{
  "repo": "OWNER/REPO",
  "start": "YYYY-MM-DD",
  "end": "YYYY-MM-DD",
  "commits": [
    {
      "sha": "string",
      "date": "ISO-8601",
      "author": { "login": "string|null", "name": "string|null" },
      "message": "full commit message",
      "summary_line": "first line of message",
      "is_merge": false
    }
  ],
  "summary_html": "<section>...</section>",
  "meta": {
    "commit_count": 0,
    "model": "string",
    "generated_at": "ISO-8601",
    "source": "github_live"
  }
}

Error Responses

• 400 – Missing or invalid dates, start > end
• 413 – Date span over 30 days
• 429 – GitHub secondary rate limit surfaced
• 502 – Upstream provider failure (GitHub/OpenAI)
• 504 – Total processing time exceeded 120 s

Frontend Notes

• Retains the original layout, palettes, and interaction patterns.
• Uses React state to handle validation, loaders, error surfaces, and empty states.
• Inserts the summary_html block with dangerouslySetInnerHTML; keep upstream prompt constrained to semantic HTML.

Testing & Deployment Tips

• Local smoke test: npm run dev, hit http://localhost:3000/deploy-report?start=YYYY-MM-DD&end=YYYY-MM-DD.
• Timeout budget: ensure repos with large commit history stay under 120 s; adjust the GitHub token’s rate limit if needed.
• Vercel deploy: the project is App Router–ready. Add the four environment variables in the project settings before deploying.

License

Internal project – all rights reserved.