← All comparisons
LemonCrow vs codebase-memory-mcp

What codebase-memory-mcp says, vs. what it scored.

Tree-sitter-based persistent knowledge graph (SQLite-backed) across 158 languages -- the most-starred tool in this comparison.

What codebase-memory-mcp says about itself
“The fastest and most efficient code intelligence engine for AI coding agents.”
“Evaluated across 31 real-world repositories: 83% answer quality, 10x fewer tokens, 2.1x fewer tool calls vs. file-by-file exploration.”
Publishes some numbers ...never against another search tool ~28.2k 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
codebase-memory-mcp 0.502 541ms 1817ms

83%/10x/2.1x numbers are real, peer-reviewed (arXiv:2603.27277), vs. raw file-by-file exploration. 0.502 MRR here 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 codebase-memory-mcp
definition 0.873 (n=1570) 0.871 (n=1570) 0.718(n=1570)
content 0.873 (n=1444) 0.864 (n=1444) 0.709(n=1444)
semantic 0.759 (n=1800) 0.576 (n=1800) 0.252(n=1800)
swebench 0.500 (n=1908) 0.493 (n=1908) 0.417(n=1908)
sessions 0.587 (n=491) 0.571 (n=491) 0.443(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 codebase-memory-mcp
astropy/astropy 0.772 0.715 0.542
atelier-ws/atelier-dev 0.467 0.477 0.344
atelier/atelier 0.594 0.557 0.415
django/django 0.689 0.652 0.406
matplotlib/matplotlib 0.801 0.747 0.578
mwaskom/seaborn 0.814 0.768 0.566
pallets/flask 0.735 0.671 0.547
psf/requests 0.840 0.803 0.637
pydata/xarray 0.815 0.764 0.636
pylint-dev/pylint 0.856 0.784 0.582
pytest-dev/pytest 0.826 0.739 0.583
scikit-learn/scikit-learn 0.740 0.669 0.463
sphinx-doc/sphinx 0.637 0.580 0.437
sympy/sympy 0.694 0.637 0.487
torvalds/linux 0.726 0.668 0.456

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, codebase-memory-mcp included. Full methodology, every raw number, and the other 9 tools →