Inkdex Markdown RAG

Example Use Cases

• 1

  Search and retrieve relevant information from markdown documentation.

• 2

  Local, zero-configuration embedding and indexing of documentation.

• 3

  Combined vector and full-text search results ranked by relevance.

Description

Inkdex indexes markdown documentation by splitting files into chunks based on headings and paragraphs, embedding them locally, and storing in SQLite. It enables efficient search over documentation using a combination of vector similarity and BM25 full-text matching, fused via Reciprocal Rank Fusion. This allows agents to retrieve relevant documentation snippets with context, improving knowledge access without external dependencies or configuration.