AI Coding Agent
An AI coding agent is an autonomous or semi-autonomous system that reads a codebase, plans changes, edits files, runs tests, and iterates toward a goal with minimal per-step human prompting.
Detailed definition
Modern AI coding agents combine a large language model with tool access (file edit, shell command, browser, MCP servers), a context window, instruction-file discovery, and a control loop. Examples include OpenAI Codex, Anthropic Claude Code, GitHub Copilot agent surfaces, and the Cursor agent. Unlike a code-completion assistant that suggests the next line, an agent can break a task into steps, observe intermediate results, and revise its approach. Reviewable safety controls — repository instructions, hooks, MCP permission boundaries, iteration caps, and human-in-the-loop checkpoints — determine whether an agent stays inside trusted boundaries while it works.