JohnXu22786/docindex

README 摘要

dsh-doc-index 中文说明 A dsh bundle that turns a workspace into a searchable local knowledge base: it builds a semantic document index over your local Markdown / plain text / PDF / DOCX / PPTX / XLSX files, and lets you (or an agent) query it with natural language or keywords — getting back the matched documents, snippets with exact line numbers and relevance scores. - Lexical retrieval: SQLite FTS5 ( node:sqlite , no native deps) with a custom CJK n-gram tokenizer for proper Chinese search. - Semantic retrieval: a provider slot with a zero-dependency local embedder by default, and an optional neural transformers provider. - Results are fused with Reciprocal Rank Fusion (RRF) . - Incremental updates: only changed files are re-indexed (with an optional file watcher), and a full rebuild is always available as a fallback. - Capacity and exclusion controls keep the index bounded and targeted. - Ships a dsh service ( ctx.docIndex ) + four model-facing tools ( doc scan , doc query , doc reindex , doc stats ) and a standalone CLI ( docindex ). The bundle follows the standard dsh distribution format: the package.json declares dsh.bundle.patch pointing at cordis.patch.yml , and the entry module…

分类