Versions:

  • 0.27.0
  • 0.26.0
  • 0.25.0
  • 0.18.0
  • 0.17.0
  • 0.11.0

Mermaid is an open-source, model-agnostic AI pair programmer designed to run in the terminal, published by Noah Sabaj. The software is currently at version 0.27.0, with six versions released in total, indicating an actively developed project that has gone through multiple iterations. Its purpose is to bring AI-assisted programming directly into the command-line environment, allowing developers to work alongside an AI coding assistant without leaving their terminal workflow. A defining characteristic of Mermaid is its model-agnostic design. Rather than being locked to a single AI provider, it works with Ollama for local model execution, as well as hosted services including Anthropic, Gemini, OpenAI, and Groq. It also supports OpenRouter and any OpenAI-compatible endpoint, giving users broad flexibility in choosing which language model powers their pairing sessions. This makes it suitable for users who prefer local inference for privacy or cost reasons, as well as those who want access to commercial frontier models. Functionally, Mermaid offers native tool calling, enabling the AI to interact with development tools as part of its assistance. It includes support for subagents, allowing delegated or specialized task handling within a session, and provides computer-use tools that extend the assistant's ability to act on the user's environment. Interaction is delivered through a clean text-based user interface (TUI), keeping the experience consistent with terminal-centric development practices. Within a software catalog, Mermaid falls into the category of AI-assisted development tools, specifically terminal-based AI coding assistants or pair programmers. Its open-source nature allows users to inspect, modify, and extend the codebase to fit their needs. Typical use cases include AI-guided code writing and editing from the command line, experimenting with multiple model providers through a single interface, local AI-assisted development via Ollama, and orchestrating multi-agent or tool-augmented coding workflows in environments where graphical interfaces are unavailable or undesirable.

Tags: