This plugin acts as a core library to integrate Large Language Model (LLM) capabilities across the odev framework. It
provides a unified, simple interface for interacting with various AI providers, allowing other plugins to easily
leverage AI for tasks like code generation, translation, and analysis.
This plugin is a foundational component and is used by other plugins such as scaffold and translate.
For the AI agent to work correctly and efficiently across all tasks (testing, upgrading, pre-commit fixes, etc.), the following tools must be installed on your host system:
- pre-commit: Essential for the AI to run and verify linting checks automatically.
pip install pre-commit
- ruff: A primary linter used by the AI agent to analyze and fix code style and quality issues.
pip install ruff
- RTK (Rust Token Killer): Critical for performance and cost. It optimizes almost all command outputs sent to
the AI (saving 60-90% on token usage).
odevautomatically binds your host'srtkbinary and its cache (~/.cache/rtk) into the AI sandbox.curl -fsSL https://raw.githubusercontent.com/rtk-ai/rtk/refs/heads/master/install.sh | sh rtk init --global
Configuration is handled automatically when you first install odev or enable the AI plugin. You will be prompted to
select your preferred LLM provider and enter the corresponding API key.
odev now automatically supports RTK to compress terminal output and reduce LLM token
consumption.
If you don't have RTK installed, you can install it using the official script:
curl -fsSL https://raw.githubusercontent.com/rtk-ai/rtk/refs/heads/master/install.sh | shTo benefit from transparent command rewriting in Claude Code, run:
rtk init --globalThis will configure a PreToolUse hook that automatically wraps your commands with rtk.