Documentation
Install GitHub

Documentation · fully offline

Teach your agent to grep less

greplm indexes your codebase locally and returns compact, jump-ready answers — so agents stop reading whole files to find one symbol.

Start the setup Browse MCP tools
  • Up to 99% fewer tokens
  • 74× faster
  • MCP-native
  • Fully offline
greplm setup live

$ curl -fsSL …/install.sh | sh

greplm installed

$ cd your-project && greplm setup

index ready · daemon warm · agents configured

$

grep + read 22,700 tokens
−97.9%
greplm search 474 tokens

Same query. same answer.

Agent playbook

Four steps from zero to token-efficient queries.

  1. 01 Install One-line script or cargo install
  2. 02 Index greplm setup in your repo
  3. 03 Connect MCP Paste config into your editor
  4. 04 Agent files Teach when to reach for greplm

Explore the docs

Guides Getting started Install, index, agent files Reference MCP server Tools & client setup Guides Usage Workflows, daemon, semantic search Guides Code intelligence Call graph, AST, context packs Guides Token efficiency Context savings & benchmarks Reference Commands Full CLI reference Reference Configuration .greplm/config.toml options More Features & comparison vs ripgrep / LSP

Also useful

Benchmarks Search and context-pack efficiency methodology Build from source Compile greplm and greplm-mcp locally