← All comparisons
LemonCrow vs fff (fff-mcp)

What fff (fff-mcp) says, vs. what it scored.

In-memory, frecency-ranked file and content search with a background watcher -- fast file discovery, not symbol-level code search.

What fff (fff-mcp) says about itself
“A file search toolkit for humans and AI agents. Really fast... Way faster than CLIs like ripgrep and fzf in any long-running process that searches more than once.”
“On a 500k-file Chromium checkout, FFF achieves sub-10ms per query compared to 3-9 seconds per ripgrep spawn.”
Publishes some numbers ...never against another search tool ~9.6k stars ↗
What it actually scored — same 14 repos, same 7,213 queries as every other tool
Tool MRR p95 p100
LemonCrow +semantic (BGE) 0.727 390ms 1057ms
LemonCrow lexical (default) 0.676 134ms 319ms
fff (fff-mcp) 0.430 46ms 207ms

Speed claim real -- 46ms p95, among the fastest measured. File finder, not symbol search: 0.430 MRR vs. LemonCrow's 0.676-0.727.

By query kind -- same benchmark, broken out (no reps in this eval: one deterministic pass per query)
Kind LemonCrow +semantic LemonCrow lexical fff (fff-mcp)
definition 0.873 (n=1570) 0.871 (n=1570) 0.684(n=1570)
content 0.873 (n=1444) 0.864 (n=1444) 0.834(n=1444)
semantic 0.759 (n=1800) 0.576 (n=1800) 0.021(n=1800)
swebench 0.500 (n=1908) 0.493 (n=1908) 0.341(n=1908)
sessions 0.587 (n=491) 0.571 (n=491) 0.281(n=491)

n = query/gold pairs of that kind, out of 7,213 total -- every provider scored on all 5 kinds.

By repo -- all 15 repos in the corpus, same query set
Repo LemonCrow +semantic LemonCrow lexical fff (fff-mcp)
astropy/astropy 0.772 0.715 0.519
atelier-ws/atelier-dev 0.467 0.477 0.269
atelier/atelier 0.594 0.557 0.284
django/django 0.689 0.652 0.444
matplotlib/matplotlib 0.801 0.747 0.495
mwaskom/seaborn 0.814 0.768 0.481
pallets/flask 0.735 0.671 0.360
psf/requests 0.840 0.803 0.511
pydata/xarray 0.815 0.764 0.505
pylint-dev/pylint 0.856 0.784 0.574
pytest-dev/pytest 0.826 0.739 0.523
scikit-learn/scikit-learn 0.740 0.669 0.370
sphinx-doc/sphinx 0.637 0.580 0.354
sympy/sympy 0.694 0.637 0.397
torvalds/linux 0.726 0.668 0.464

MRR per repo: n-weighted blend across all 5 query kinds, same 7,213-query run.

The true story

Same 14 repositories, same 7,213 query/gold pairs as every tool here, fff (fff-mcp) included. Full methodology, every raw number, and the other 9 tools →