Show HN: Kbot – terminal AI agent that learns from every user who uses it (github.com)

The recently announced Kbot is an open-source terminal AI agent designed to operate independently of specific providers and models, a departure from existing solutions that often limit user flexibility. Kbot supports 20 AI providers, including well-known names like GPT and Claude, allowing users to switch between them effortlessly with a single command. It operates entirely offline using embedded models, ensuring privacy and no costs, making it a significant player in the AI/ML landscape by promoting a local-first approach. Kbot features 23 specialized agents capable of handling various tasks—ranging from code debugging to research—by intelligently routing requests to the most suitable expert agent. With access to 290 integrated tools targeted at diverse applications such as web development, game design, and database management, Kbot not only enhances productivity but also adapits to user habits over time through its Bayesian learning mechanism. Additionally, it provides a programmatic SDK for developers to integrate Kbot directly into their applications, alongside a built-in MCP server that connects seamlessly with popular IDEs. This robust functionality positions Kbot as an innovative tool in the AI ecosystem, emphasizing user-centric design and adaptability.