Retrieve upstream implementation evidence before coding with XERJ
Build a focused reference corpus, retrieve implementation passages with file-and-line citations, and check license constraints before adapting unfamiliar code.
Prerequisites
XERJ binary; Git; reachable XERJ node; shell-capable agent or configured MCP client
Installation
Install or set up from the source-backed instructions:
Use the upstream v1.0.0-rc.100 release or official installation instructions to install the platform binary; the shell installer places it in ~/.local/bin. On a trusted single-user dev machine, start xerj --insecure --data-dir ~/xerj-data and wait for the loopback node at http://localhost:9200. This disables TLS/auth; use the documented authenticated configuration on shared hosts. Keep storage outside indexed directories. Add the binary directory to PATH if required. Then run xerj corpus add reference-task ..., xerj corpus index reference-task, and xerj code reference-task "". No npm dependency is asserted. For MCP, configure the absolute binary path with argument mcp and the appropriate XERJ_URL; this proxy requires an already running node.