Droid is an enterprise-grade AI coding agent from Factory that operates in the terminal to handle end-to-end software development workflows including feature development, refactoring, bug fixes, and documentation. It solves the problem of fragmented AI tooling by providing a single command-line agent that supports multiple AI models from different providers, letting developers switch between Anthropic and OpenAI models without changing tools. Droid achieved the top score of 58.75 percent on Terminal-Bench, setting the state-of-the-art benchmark for terminal-based coding agents.
Droid features specialized agent types including Code Droid for engineering tasks like feature implementation, refactoring, and bug fixes, and Knowledge Droid for research and documentation that searches across your codebase, documentation, and the internet to answer complex questions and generate specifications. It supports Model Context Protocol for connecting to external tools, subagents for specialized tasks, and slash commands as shortcuts for common workflows. Droid's multi-model support makes each model work better by optimizing prompts and interactions for the specific capabilities of each provider.
Droid is designed for professional developers and enterprise engineering teams who want a terminal-based AI coding agent with the flexibility to use different AI models for different tasks. It is particularly appealing for teams that have already adopted terminal-centric workflows with tools like Claude Code or Codex CLI and want to compare or combine models from different providers in a single interface. Compared to other CLI-based agents, Droid's specialized agent types and state-of-the-art Terminal-Bench performance make it a compelling choice for teams that prioritize autonomy and reliability in their AI coding workflows.