What is Waku used for?

Waku is used to manage multiple local AI coding agent CLIs from a single native desktop app. It brings sessions, transcripts, tool activity, and usage into a unified timeline so developers can track work across different agents.

Which AI coding agents does Waku support?

The provided information lists support for Amp, Claude Code, Codex CLI, Cursor CLI, Grok Build, OpenCode, and Pi. Waku is described as driving these tools over their native protocols with auto-detection.

Does Waku store data locally?

According to the provided information, Waku keeps projects, sessions, transcripts, and Git-backed checkpoints on the user's machine. It is also described as requiring no account and having no telemetry, but these privacy-related claims should be verified against the official documentation.

How does Waku's Git-backed rewind work?

Waku is described as creating a Git-backed checkpoint for every prompt. This allows developers to roll back code changes and the associated conversation context together when an agent's output needs to be reviewed or reverted.

The provided tool information lists Waku as free. No additional pricing details, paid plans, or usage limits were included, so users should confirm current pricing on the official Waku website.

Who should consider using Waku?

Waku is a strong fit for developers who already use multiple AI coding agent CLIs and want one keyboard-first place to manage them. It is especially relevant for local-first workflows where session history, transcripts, and Git checkpoints are important.

Last updated: 10/1/2026Last verified: 2026-08-14

Waku Review: Local Desktop Manager for AI Coding Agents

13views

0saved

Waku is a fast native Rust/GPUI desktop app that manages all your local coding agents — Claude Code, Codex, Cursor CLI, OpenCode, Amp, Grok Build, and Pi — in one keyboard-first timeline. Projects, sessions, transcripts, and Git-backed checkpoints stay entirely on your machine, with no account and no telemetry.

Free

What is Waku?

Waku is a native desktop app built with Rust and GPUI for managing local AI coding agents from one interface. It is designed to work with agent CLIs such as Claude Code, Codex CLI, Cursor CLI, OpenCode, Amp, Grok Build, and Pi. Waku focuses on a keyboard-first workflow with local projects, sessions, transcripts, and Git-backed checkpoints stored on the user's machine.

Key Features

Manages multiple local AI coding agent CLIs, including Amp, Claude Code, Codex CLI, Cursor CLI, Grok Build, OpenCode, and Pi
Uses auto-detection and native protocols to connect with supported agent CLIs
Provides a unified timeline for sessions, transcripts, tool activity, and usage across different agents
Creates Git-backed checkpoints for prompts so developers can rewind both code changes and related conversation context
Offers keyboard-first controls for actions such as queueing, steering, and stopping agent work
Stores projects, sessions, transcripts, and checkpoints locally with no account requirement or telemetry stated in the provided information

Best For

Developers who use more than one AI coding agent CLI and want a unified local workspaceTeams or individuals who prefer keyboard-driven workflows for coding agent managementProgrammers who want to track agent sessions, transcripts, and code changes togetherUsers who want Git-backed checkpoints when experimenting with AI-generated code changesDevelopers who prefer local-first tooling without creating another hosted service account

Pricing

Waku is listed with a free pricing model in the provided tool information. No paid tiers, usage limits, enterprise plans, or optional add-ons were provided, so the official website should be checked for the latest pricing details before making a decision.

Pros & Cons

Pros

Centralizes several AI coding agent CLIs in one desktop timeline
Local-first design may appeal to developers who want projects and transcripts kept on their machine
Git-backed rewind can help reduce friction when reviewing or reverting agent-driven code changes
Keyboard-first controls are useful for developers who prefer fast, command-oriented workflows
No account requirement is stated in the provided information

Cons

Waku appears to depend on agent CLIs the user already has installed or owns
It may not replace full AI coding environments for users who prefer an integrated IDE experience
Supported agents are limited to the CLIs listed in the provided information
No details were provided about operating system availability, installation requirements, or collaboration features
No information was provided about support options, update cadence, or commercial licensing

Alternatives

Cursor

Cursor is an AI-focused code editor that integrates coding assistance directly into the development environment rather than managing multiple local agent CLIs from a separate desktop timeline.

Claude Code

Claude Code is a coding agent CLI for working with Anthropic's Claude models in developer workflows, and it is one of the agent tools Waku is designed to manage.

OpenCode

OpenCode is an AI coding agent option for terminal-based development workflows and is relevant for users comparing local coding-agent tools.

Codex CLI

Codex CLI is a command-line AI coding tool for developers who want agent-style assistance directly from the terminal.

Aider

Aider is an AI pair-programming tool that works with local Git repositories and is a practical alternative for developers focused on terminal-based code editing with version control.

FAQ

View all Waku alternatives →

Details

Platform

macOSLinux

Features

Drives the agent CLIs you already own (Amp, Claude Code, Codex CLI, Cursor CLI, Grok Build, OpenCode, Pi) over their native protocols with auto-detection
Unified timeline that normalizes sessions, transcripts, tool activity, and usage from every agent into one view
Git-backed rewind: every prompt checkpoints the working tree so you can roll back code and conversation together
Fast GPUI-rendered native window with keyboard-first controls (queue, steer, stop) and fully local storage — no account, no telemetry

Languages

Known limitations

Prebuilt downloads currently target Apple Silicon; Linux requires building from source and Windows is still planned
Embedded browser and experimental computer-use integration remain macOS-only for now
Young project (pre-1.0) that evolves quickly
Needs at least one supported agent CLI installed and authenticated before anything works