Skip to main content
MCP server CX Codex

XERJ — Claude Code MCP server: search and setup

XERJ is a local MCP server that indexes code, docs, specs, and logs, returning cited answers for project search and agent memory.

Apache-2.0 ★ 19,252 updated 2026-10-29

Open on GitHub

What XERJ does

XERJ runs as a single Rust binary and exposes a local MCP server. It indexes source code, documents, specifications, logs, JSONL, CSV, and PDF files, then searches them with BM25, vector, or hybrid retrieval. Results include file and line references.

It also supports code lookup, agent memory under /_memory, and an Elasticsearch-compatible API. Run xerj init to connect XERJ to Claude Code or Cursor through MCP and a skill; xerj autoindex ~/my-project starts indexing a project.

Keep the data directory outside any tree that will be indexed, since data/ is not ignored by default. The --insecure flag disables TLS and API-key authentication, so any client that can reach the port can read or delete indexes.

What it helps with

  • search project code and documentation
  • index source files, logs, and PDFs
  • locate function definitions with file and line references
  • store agent memory in /_memory
  • run several searches through xerj_batch_search

Example prompts for the agent

  • Use XERJ to index this project and find references to WAL rotation.
  • Find the definition of euler_to_rotationmatrix and return its file and line.
  • Run the questions from queries.json through xerj_batch_search.

How to install XERJ

Install

The MCP registry has no launch command for it: see the repository README

Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more

Questions about XERJ

How do I install XERJ in Claude Code?

Run curl -fsSL https://xerj.org/get | sh, start xerj --insecure --data-dir ~/xerj-data &, then run xerj init. The last command connects XERJ to Claude Code through MCP and a skill.

What does XERJ do?

It builds a local index of code, documents, and logs, then returns matching passages with citations for an AI agent. Search supports full text, vectors, and hybrid retrieval.

Where should the XERJ data directory go?

Place it outside the folder tree you plan to index. With --data-dir ./data inside that tree, XERJ can index its own storage.

Similar MCP servers

MCP servers catalog →

Project sponsors

Companies that keep this analytics open