Graqle

Our take

Python package (pip graqle, ~700 weekly downloads, v0.76) from Quantamix Solutions that scans a codebase into a typed knowledge graph via AST and dependency analysis, then serves graph-powered reasoning — impact analysis, multi-agent audits with confidence scores, persisted 'lessons' — to MCP clients, plus a separate runtime layer that signs production AI decisions and anchors them to the Sigstore Rekor transparency log. Actively developed (pushed June 2026) but the README leans hard on unverifiable marketing (EU AI Act alignment, a case study claiming 53-88% token-cost savings, 'patent-defensible'), and the license is Apache-2.0 text with a patent rider forbidding reimplementation or extraction of the methods (three pending EP filings) — so despite the PyPI 'Apache-2.0' label it is not straightforwardly open source, which is why GitHub reports NOASSERTION. For teams wanting architecture-aware code reasoning who accept a single-vendor, patent-encumbered stack; everyone else should read the LICENSE first.

reviewed by hand · 2026-07-05

Something wrong or dead here? Report it

Verification record

last verified

29 days ago

github stars

32

last commit

2 mo ago

archived

no

license

NOASSERTION

in registry since

2026-03-16

github.com/quantamixsol/graqle