Hey HN, I built Oruk, a filterable breaking news wire that you can consume via web, API, or terminal. What's cool is it's monitoring speech around the world (just radio stations for now) and transcribing intt in hyper-low latency stories.

I kept running into the same problem building side projects: getting reliable, structured access to live news data is surprisingly painful. Most news APIs are expensive, slow, or return garbage. I wanted something I could pipe into scripts, feed into LLM workflows, or just browse quickly without a bloated interface.

Oruk aggregates breaking events globally and organizes them by time, region, and category. On the technical side, we support SSE streams for continuous consumption, a REST API with tiered quotas, a CLI tool, and MCP integration for agent workflows. API key management is built in for teams.

The web frontend is just a thin client on top of the same API anyone can access.

Would love feedback on the API design, the data quality, and anything that feels broken.