We finally put in place a solid process to test our agents with evals. Agents can now be evaluated against a dataset after prompt or model changes. Production traces can be easily added to the dataset. And humans can participate to sample and recalibrate the evals when needed.

Is there a way to have custom modes versioned in a git repo with cursor?

New cover!

This is an arrangement of Schala's Theme from Chrono Trigger arranged entirely for metallic and glassy percussion - featuring the members of Open World Percussion!

Check out the full album "Chambers of Time" by GameGrooves!

Info below! 🔽🔽🔽

Full video on YT: youtu.be/-hc_87LCDEA?...

One tech recruiter asked me for a minimum of 5 years of experience with LangChain. I pointed out that LangChain was created only 2 years ago. He did not believe me; said he already had other candidates with 4+ years of doing LangChain 🤦‍♂️

If you wanted to experiment with a POC for creating an Agent, what would your stack be? LangChain?

I'm honestly very happy to have been pushing SST at Prismic. Infrastructure as Code is key for a company of our size. Also, the fact that we're now using TypeScript for the frontend, backend, and infrastructure is definitely a game-changer!

Decided to start one of these, inspired by @simonwillison.net.

til.samwho.dev

Show HN: Pretty-print your chess games using Prettier Discussion

I propose we replace semantic versioning with pride versioning

Funny, Just reach limit of number of interactions you can have with Cursor Agent!