Oko: code search for coding agents
Oko uses Jev to rank code for agents. Its author reports 0.48 vs BM25's 0.21 for needed code within 8k tokens, and faster Claude Code sessions across nine tasks.
# Oko **Help your coding agent find code faster and use fewer tokens.** Oko helps Codex, Claude Code, and OpenCode spend less time searching and fewer tokens reading irrelevant code. It delivers relevant source snippets through MCP so your agent can get to the task sooner. Gains vary by task and coding tool. On [Agent Retrieval Bench](#retrieval-quality), a public benchmark of 345 code-retrieval tasks in six languages, Oko puts a right file first more often than any published method (MRR 0.39 against 0.24), with no GPU and no index. In [our own agent benchmark](#benchmarks), Codex, OpenCode, and Claude Code use 13–44% fewer tokens and 33–61% fewer tool calls with it; OpenCode and Claude Code also finish tasks 21–23% faster. Oko runs locally and uses [TypeSafe AI’s Jev](https://typesafe.ai/) to rank selected source snippets. Use it through your coding agent or directly from your terminal. [Install](#install) · [Connect your coding tool](#connect-your-coding-tool) · [Benchmarks](#benchmarks) · [CLI examples](#use-in-your-terminal) · [Documentation](#documentation) ```sh oko ask "where is authentication handled?" ``` No Node.js or Rust runtime needed for downloaded binaries. Oko needs a TypeSafe AI API key for Jev ranking; local keyword search also works without a key. ## Install Run this on macOS or Linux to install the current prerelease: ```sh curl -fsSL https://raw.githubusercontent.com/bartlomein/oko/main/install.sh | sh export PATH="$HOME/.local/bin:$PATH" ``` The installer detects your processor, downloads Oko and its bundled ripgrep, verifies the checksum, and installs them without administrator access. Add the `export PATH` line to `~/.zshrc` or `~/.bashrc` to keep `oko` available in new terminals. Existing installations from other sources are left unto