wtf is confluence2md-indexer?

confluence2md-indexer takes documents exported from Confluence, a workplace wiki tool, and turns them into a searchable local database on your own computer. It is the second step in a three-part pipeline: first confluence2md exports Confluence pages to markdown files, then this tool builds a search index from those files, and a separate tool called confluence2md-mcp lets AI assistants query that index. Once you point it at an export folder, it reads the markdown pages, breaks them into smaller chunks of text, and stores those chunks in a single SQLite database file along with their embeddings, which are numeric representations used for finding similar meaning rather than just matching exact words. From the command line you can then search that database three ways: plain keyword search, meaning-based vector search, or a hybrid of both, with results returned as readable text or as structured JSON. The tool re-indexes only what changed since the last run by default, though a full rebuild is available when needed, and it removes records for documents that no longer exist in the source export. It also exposes a Go API so another program, such as the MCP connector, can query the index directly without running the command-line tool as a separate process. The README notes this is built for local, single-developer use rather than serving many users at once, and includes guidance for estimating how large the resulting database will get based on how much markdown text you feed it. Using it requires Go 1.25 or newer to build from source, though pre-built binaries are also offered on the project's releases page. It is licensed under MIT.