Recently, I noticed my debugging workflow had become: copy error into ChatGPT/Claude -> go back and repeat (while being cautious not copying over sensitive stuff like paths, names etc.)

So, I built Sleuther, a Oh My Zsh plugin that asks a local LLM via Ollama to explain and debug errors directly in the terminal. Right now I am using Qwen2.5-Coder.

At a previous job, I used warp.dev extensively and was missing its agent mode. Sleuther is inspired by warp and zsh-copilot.

No copy paste, no cloud, no context switching.

Also, has a small trick: if the local LLM cannot help after a couple attempts, it suggests a sanitized version of the error, stripping usernames, paths, and hostnames before sharing with ChatGPT or Claude. The sanitization is rather rudimentary right now but something I want to improve.

Curious if anyone else is experimenting with local LLMs for dev workflows?

https://github.com/chocks/sleuther