Same experience, here. Very hard to base on facts, because every problem and prompt is an individual use-case and measuring agent reasoning quality is notoriously difficult anyway. But I spend a lot of time with Claude and my overall "feeling" fully matches your description. Quality has deteriorated, thinking takes longer and results become shallow. Something is off...

- Compile a resulting set of environment variables from a hierarchical list of `<name>.env` files. - Dependencies can form a tree or DAG (directed acyclic graph). - Smart environment selection via builtin FZF (fuzzy find). - Side-by-side edit of trees for easy comparison. - [direnv](https://direnv.net/) integration - [JetBrains](https://www.jetbrains.com/) integration via [EnvFile](https://plugins.jetbrains.com/plugin/7861-envfile) plugin.

Rust based, fuzzy find, full text search, tags, CLI only: https://github.com/sysid/bkmr

Super fast, full text search, fuzzy search... For command line dwellers: https://github.com/sysid/bkmr

For starlette/fastapi there is a battle tested lib: https://github.com/sysid/sse-starlette

I4 and BMW as trigger to investigate Truchet tiles.

We all need to cope with Corona. What is your strategy being grounded at home?