andymik90/aperant

Auto Claude is a desktop application that runs AI coding agents on your software project autonomously. Instead of asking an AI assistant individual questions, you describe a task, and the system plans the work, writes the code, and checks its own output, cycling through that loop until the task is done or flagged for your review. The interface centers on a Kanban board where you create tasks and watch agents move them through stages. Behind the scenes, each task spins up one or more agents that work inside isolated copies of your codebase called git worktrees, so your main working branch is never touched until you approve the result. You can run up to twelve agents at the same time on different tasks, and a built-in quality-assurance step runs before any result is presented to you. When multiple agent results need to be merged back into the main branch, the app handles conflict resolution automatically. Agents retain memory across sessions, meaning insights from one task carry forward to the next so work builds on itself rather than starting from scratch each time. The app also integrates with GitHub and GitLab, letting you import existing issues and have agents investigate and address them, then create merge requests. There is also a Linear integration for teams that track work there. Using the app requires a Claude Pro or Max subscription and the Claude Code command-line tool installed on your machine. Your project also needs to be a git repository. Desktop installers are available for Windows, macOS, and Linux. A command-line mode exists for running agents in headless environments or in automated build pipelines. The project is open source under the AGPL-3.0 license. The README notes that all releases are scanned with VirusTotal and include checksum files for verification.