News for hackers. Stuff that matters.

alvarogar

The main reason we've seen is a bad first impression: the agent misunderstands the initial prompt, doesn't load the available tools or starts doing something clearly wrong, and the user ctrl+c's instead of correcting it since it's faster to just start a new one.

The learning is that it is fixable. Better CLAUDE md instructions, clearer initial prompts, and skill configurations that reduce the uncertainty cut abandonment significantly in our team.

Show HN: Run SigNoz on ObsessionDB and ClickHouse Cloud

https://github.com/obsessiondb/signoz-obsessiondb

Hey HN, I'm Alvaro. I work on ObsessionDB, a managed ClickHouse service. We run petabyte-scale ClickHouse infra as a service.We open-sourced a fork of SigNoz's schema migrator that makes SigNoz work on SharedMergeTree. That means it runs on both ObsessionDB and ClickHouse Cloud.The problem: SigNoz's production setup assumes sharded ClickHouse. The official docs spec 56 CPU cores, 152 GiB RAM...

( Read More... | no comments | 2 points | on HN )

Show HN: Open-source ClickHouse skills – SQL craft course for your agent

https://github.com/obsessiondb/clickhouse-skills

Hey HN, I am Álvaro and I am working on ObsessionDB, a managed ClickHouse infra as a service. We run CH at about 100B rows scale and we spent the last few years learning what works and what doesn't at that scale.These are open-source skills that teach Claude Code, Cursor, or any agent the ClickHouse patterns that matter.The problem is that AI agents write syntactically correct ClickHouse SQL,...

( Read More... | no comments | 2 points | on HN )