FireGen: Turn Firebase RTDB into Your Universal Generative AI API (wechaty.js.org)

FireGen is an open-source Firebase Extension (MIT) that converts Firebase Realtime Database (RTDB) into a universal generative-AI job queue, announced as a hackathon MVP aimed at solo founders and AI-native developers. Instead of wrestling with Vertex AI SDKs, long-running operation (LRO) polling, GCS file juggling, and auth, you push a job node (or just a string) to /firegen-jobs and subscribe via RTDB’s onValue — FireGen handles model selection, polling, retries, storage, signed URLs, and auth using Cloud Functions v2 and Firebase Task Queues. The project compresses days of integration work into minutes and emphasizes zero-config defaults and low-latency RTDB listeners for simple, real-time status updates. Technically, FireGen offers two modes: an AI-Assisted mode where a Gemini 2.5 Flash–powered Request Analyzer semantically interprets plain-text prompts and picks models like Veo 3.1 (video), Gemini Image, or TTS while logging its reasoning to the DB; and an Explicit mode that mirrors the Vertex AI REST API for production control. It forgoes bloated SDKs for direct REST calls, uses TypeScript/Node.js 22 with Zod validation, and integrates GCS signed URLs for large files. The result: a developer-friendly, agent-ready interface that abstracts LRO complexity and accelerates prototyping and AI-to-AI workflows, letting teams focus on product rather than plumbing.