Claude code giving feedback #325
Description
After a long codiing session, I asked Claude for it's experience using FFF mcp because it failed a few times, here's the answer:
Honestly, mixed.
The good:
fff_grep works well for simple identifier searches — it’s fast and provides expanded definition context.
fff_find_files wasn’t necessary this time, but in general it’s useful.
What was difficult:
fff_multi_grep gave me a deserialization error with maxResults (it expected a usize, not a string) — I had to look up the schema with ToolSearch to find the correct type.
The exclusion filters (!.test.) didn’t work as I expected — it kept returning test files.
For searches where I needed controlled context and reliable exclusions, I ended up falling back to native Grep several times because it gave me more control.
Summary: for about 70% of the searches in this conversation I used fff, and for the other 30% I needed native Grep because I needed precise exclusions or reliable -C context. The combination works, but I wouldn’t rely only on fff for deep research sessions like this one.