Agents Daily
Curious about
I read the Cloudflare changelog, model-lab blogs and agent-framework releases every day anyway. I wanted that reading to land somewhere public, with a link back to every source, without me writing it by hand.
Built
One Worker on daily.rclabs.in. A cron at 08:15 Berlin time pulls official feeds (Cloudflare blog and changelogs, OpenAI, Google, DeepMind, Anthropic, Hugging Face, Mistral, and releases from agent frameworks like LangGraph, CrewAI and the Agents SDK), dedupes by URL, and stores items in D1. A free Workers AI model summarises each item from the source text only, then the day is built into a digest.
Every story and every day gets its own page, with tag pages, weekly best-of pages, RSS, a sitemap and structured data. There is one sponsor slot per digest, and the sponsor page shows real traffic numbers.
Learned
- Some feeds publish only a title. A summariser fed a title will invent the rest, so those stories get a short fallback instead of a made-up paragraph.
- Release notes summarise better than blog posts. They’re already lists of changes.
- Search traffic takes weeks. Per-story pages are what it eventually finds.
next specimen → EXP-002 · Home HQ