Agents & tools
Coding Agent
An AI agent specialised for writing, editing, and debugging code — usually with the ability to read your repo, run tests, and open pull requests.
01 ——
In plain English
A coding agent is an AI agent whose primary tools are code editors, terminals, and version control. Where an autocomplete suggests, a coding agent acts — it can navigate a codebase, edit multiple files, run tests, and submit a PR with no human in the middle.
What distinguishes coding agents from autocomplete:
- Multi-file edits — refactors across the whole repo, not just the current line
- Tool use — runs the test suite, lints, type-checks, git operations
- Long horizon — completes a feature or fixes a bug end-to-end
- Self-correction — sees the test fail, edits again, runs again
Leading products (early 2026):
- Cursor and Windsurf — agent-mode IDEs
- Claude Code and Codex — terminal-based agents
- Aider — open-source pair-programmer
- Devin (Cognition) — fully autonomous engineering agent
- GitHub Copilot Agent Mode — in-IDE agent from GitHub
The frontier is shifting from "agent suggests" to "agent does, human reviews PR."
02 ——
Related terms
AI Agent
Software that uses an LLM to plan and act — picking tools, taking actions, and adapting based on results to complete a user’s goal.
Agentic AI
AI systems designed to act, not just respond — they plan, use tools, and make decisions across multiple steps to complete a goal.
Copilot
An AI assistant embedded directly into a workflow — like coding, writing, or design — that suggests, completes, or generates work alongside the user.
Code Generation
Using an AI model to write source code from a natural-language description, a partial snippet, or a test.
Agent Harness
The scaffolding around an LLM — tools, memory, loops, and orchestration — that turns a model into an agent.
Vibe Coding
A 2025–26 coined term for writing software by chatting with an AI agent — describing the vibe of what you want and accepting whatever code it generates.
Back to glossaryLast reviewed May 2026