wsmoak.bsky.social

@wsmoak.bsky.social

W
W

Claude is figuring out how to integrate LangChain's OpenSWE with DevPod as the backend for sandboxes. It is slowly making progress... There is _zero_ chance any of this would be happening if I had to read all the code and docs and figure it all out by myself.

Claude session showing it using subagents to build, push, deploy and watch the task status before it proceeds with commenting on a GitHub issue to trigger a webhook.

Having a linguistics background I think there's a huge misconception when it comes to prompting LLM's, the vast majority thinks that coding is "hard" compared with speaking a natural language. Well it's actually the contrary from a utilitarian POV less verbs at your disposal => easier to be specific

Kelsey Hightower@kelseyhightower.com · 6mo ago

Generative AI is a slop generation machine by default. You have to put in a lot of work to get something of quality from it.

W

Open Source in 2025: clone a git repo, discover it's in a language I've never used, install the language, build the project, try the examples, realize it's missing a feature I want, have Claude Code implement the missing feature. So now I guess I get to learn Rust and see if the code is correct!

There are so many cranes on the Platte that it can be a fight for space—as one flock takes flight, another fills in their spot. On our last morning on the river, researchers conducting aerial surveys estimated an all-time record of 736,000 Sandhill Cranes on the 80-mile stretch of the Platte River.

Bild
W

Anthropic's Model Context Protocol + #Claude Desktop is another level. In less than an hour I had Claude write an #MCP server for retrieving data from the YNAB (You Need a Budget) API. Now I can ask it about account balances and spending by category. It just works.