xerj-org/xerj — 7,066 Stars
XERJ documents code and document search with autoindex and MCP. Footage shows the author playground with MOCK DATA · BACKEND PENDING and README commands.
Watch Episode
About This Repo#
xerj-org/xerj — 7,066 ⭐ | Rust
XERJ documents an autoindex workflow for source files, documents, logs and PDFs. Its code results include symbols and line numbers, and agents can retrieve relevant context over MCP. Autoindex creates one index per dataset, not necessarily one index per folder.
Footage and Source Scope#
Footage source: the author’s public playground https://xerj.org/playground/#/discover displays MOCK DATA · BACKEND PENDING. The opening shows interactions with that author-provided mock example; it is not an independent live search-engine test. The remaining footage shows native-dark GitHub README commands and documented workflow, not local runtime execution.
No universal token-savings claim is made here. Optional reranking, proxy embeddings and WAL export can send data off-node when configured; see the security model before choosing a deployment.
Narration#
Can’t find where your project explains something? Xerj makes code and documents searchable together. Point autoindex at a folder containing source files, logs, or PDFs, then search for the passage you need. Code results include symbols and line numbers. Connect your coding agent over M C P to retrieve relevant context.