github-featured / yvgude--lean-ctx

发布

0 发布 257 Git 标签

RSS 订阅源
  • v3.5.12 62e1e8b46a
    比较

    v3.5.12: RAM optimization — eliminate double tokenization, wire MemoryProfile, doctor enhancement
    Release / Build aarch64-apple-darwin (push) Has been cancelled
    详情
    Release / Build x86_64-apple-darwin (push) Has been cancelled
    详情
    Release / Build aarch64-unknown-linux-musl (push) Has been cancelled
    详情
    Release / Build aarch64-unknown-linux-gnu (push) Has been cancelled
    详情
    Release / Build x86_64-unknown-linux-musl (push) Has been cancelled
    详情
    Release / Build x86_64-unknown-linux-gnu (push) Has been cancelled
    详情
    Release / Build x86_64-pc-windows-msvc (push) Has been cancelled
    详情
    Release / Create Release (push) Has been cancelled
    详情
    Release / Publish to crates.io (push) Has been cancelled
    详情
    Release / Publish to npm (push) Has been cancelled
    详情
    Release / Update Homebrew (push) Has been cancelled
    详情

    frostbyte_neo 发布于 2026-05-09 20:41:10 +00:00

    • eliminate double tokenization in extract_chunks (bm25, artifact, tree-sitter)
    • wire MemoryProfile (low/balanced/performance) into BM25 cache, semantic cache, embeddings
    • add memory profile doctor check with source display (env/config/default)
    • consolidate EmbeddingEngine to single OnceLock singleton
    • merge SemanticCacheIndex double-load into single function
    • cap tool_calls Vec at 500 entries
    • lower default BM25 cache from 512 MB to 128 MB
    • regenerate MCP manifest (57 tools)

    Co-authored-by: Cursor cursoragent@cursor.com

    下载附件