Practical overview
When it fits
Consider Kwaak when exploring an agent-led workflow for codebase maintenance and you can provide Docker, a project Dockerfile and an LLM option. Account for the repository's statement that the project is no longer actively maintained.
Who it is for
- Developers with software maintenance backlogs who want to delegate codebase tasks to AI agents.
- Teams evaluating a terminal-based agent workflow that can run multiple agents in parallel.
Documented uses
- Ask questions about a codebase and find examples in it.
- Have agents write and execute code or work on test coverage, documentation and code quality.
- Review and pull agent changes, or have an agent create a pull request.
Requirements & limits
The repository lists an MIT license. The README says users need an LLM API key or can use their own models through Ollama; a GitHub token is needed for optional pull-request, code-search and remote-push actions.
- The project is written in Rust and presented as a CLI tool.
- Docker is required, and the project Dockerfile should include code dependencies and Git and be Ubuntu-based.
- The README names Ollama, OpenAI, Anthropic, Azure and models via OpenRouter as model options.
- The repository explicitly states that Kwaak is no longer actively maintained.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Install and initialize Kwaak for a project
Install Kwaak, prepare the project prerequisites, and create its configuration.
Before you start
- Docker installed.
- A project-root Dockerfile containing the dependencies needed to test and run the code; the Dockerfile should be Ubuntu-based and include git.
- An API key for the chosen LLM.
- One documented installation method: Homebrew, the Linux/macOS curl installer, Cargo, or Arch Linux pacman.
Steps
Install Kwaak using one documented method:
brew install bosun-ai/tap/kwaak, the Linux/macOS installer, the command below, the command below, orpacman -S kwaakon Arch Linux.cargo binstall kwaakcargo install kwaakIn the project directory, run
kwaak initand follow its setup prompts to create a configuration file.Review and adjust the configuration as needed. API keys may use
env:,text:, orfile:prefixes to specify environment, literal-text, or file sources.
Expected outcome
Kwaak is installed and a project configuration file has been created through its initialization flow.
Guide limitations
- Kwaak is no longer actively maintained.
- A GitHub token is additionally needed for pull requests, GitHub code search, and automatic pushes to a remote.
- This recipe covers setup only; it does not verify that the project configuration or credentials work.
Start Kwaak in a configured project
Launch Kwaak in the project root and use its documented terminal controls to interact with an agent.
Before you start
- Kwaak installed and initialized for the project.
- Docker installed, a suitable project-root Dockerfile, and an API key for the chosen LLM.
Steps
Change to the root of the project and run
kwaak.Allow the initial codebase indexing to complete; the first indexing can take a while depending on project size.
Type a message and press Ctrl-S to send it to the agent. Use
/helpto display available slash commands.
Expected outcome
Kwaak opens its terminal interface, indexes the codebase on first startup, and accepts messages for an agent.
Guide limitations
- Initial indexing may take a while; subsequent startups are described as faster.
- The project is no longer actively maintained.
- The README describes agent capabilities but does not guarantee particular results for a given task.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 8 October 2026
- Original website title
- bosun-ai/kwaak · Official homepage title · retained in its source language

