christopherkarani/Wax
Single-file memory layer for AI agents, sub mili-second RAG on Apple Silicon. Metal Optimized On-Device. No Server. No API. One File. Pure Swift
ai-agentsclicoremlcoreml-frameworkdata-sciencemachine-learningmcpmcp-servermemorymemory-cachememory-hackingmetalon-device-airagrag-pipelineswiftvector-databasevector-embeddingsvector-searchvectordb
First Claude commit: Feb 16, 2026Last Claude commit: 1mo agoDiscovered: Mar 24, 2026
Recent Claude Commits
Fix all production-readiness audit findings across 22 files
763d2991mo agoauthor_emailCorrect two false positives and downgrade one finding after source verification
546ed931mo agoauthor_emailExpand test coverage section with agent findings on missing test areas
d4576e31mo agoauthor_emailaudit: add CoreML force unwrap and BertTokenizer lock findings
72d5e251mo agoauthor_emailaudit: add CLI credential leakage and path traversal findings
2df6bdf1mo agoauthor_emailaudit: add FTS5Serializer uninitialized memory finding, fix numbering
d2b20f41mo agoauthor_emailaudit: add Metal/GPU findings from WaxVectorSearch deep dive
1f675271mo agoauthor_emailaudit: comprehensive production-readiness report for Wax framework
c7918ff1mo agoauthor_emailFix 30-second demo crash: auto-fallback to text-only when no embedder provided
52a0cb72mo agoauthor_emailFix MCP server bugs: photo init waste, malformed JSON, unsafe arithmetic, signal handling
f4d16402mo agoauthor_emailAdd deep analysis of Wax MCP server implementation
facb8b92mo agoauthor_emailrefactor: use opaque `some` types instead of existential `any` in function parameters
de586742mo agoauthor_email