Show HN: I made clawfeeds, feeds for agents
Clawfeeds is an RSS feed parser designed for AI agents that converts feeds into context-efficient markdown files with webhook notifications for new posts, optimizing token usage for LLM-based agents.
You can try it by simply asking Claude to check out https://agent.clawfeeds.com/try.md, it'll figure it out, or have a look yourself at https://clawfeeds.com
- (very) context-friendly feeds (feeds & posts separated out) - webhook for new post notifications
It crawls RSS feeds, parses them into markdown, and gives you a very concise feed markdown file (example: https://agent.clawfeeds.com/feed/uqhb-tssd-4gk9.md), which links to markdown post files (example: https://agent.clawfeeds.com/post/cm47-5dbd-x6we.md).
The idea is 1. protect your context window, and 2. get notified of new posts at your webhook (or some other mechanism).
It's free - it doesn't eat up much resources, the markdown files live in a DO bucket. You need an account to add/follow feeds and add a webhook for your agent to be notified of new posts from feeds you follow.
I can think of a ton more features (it's kind of a feed reader for agents), but I'd love to hear what people think might be useful.
An interesting aspect has been to design this for agents first - the CTA is aimed at agents, etc.
I imagine use cases like having my agent read my favourite Substacks and summarizing things, following packages I use on Github etc. RSS is (still) everywhere. (For example this https://news.ycombinator.com/item?id=47337607 recent discussion)