What is AI Agent?
AI Agent — A reusable set of instructions that gives an AI tool a specific persona, expertise, and workflow.
What is an AI agent?
An AI agent is a reusable set of instructions — typically written in Markdown — that gives an AI assistant a specific persona, domain expertise, and step-by-step workflow. Unlike a one-off prompt, an agent persists across conversations and projects, producing consistent, high-quality results every time it's used.
How do AI agents work?
When you load an agent into an AI coding tool like Claude Code, Cursor, or Windsurf, the agent's instructions become the AI's operating context. The AI assumes the defined persona, follows the specified workflow, and applies the rules and constraints the agent defines.
For example, a Code Reviewer agent doesn't just read your code — it follows a structured process: understand context, check correctness, assess security, evaluate performance, and provide severity-rated feedback with code examples.
What does an AI agent definition include?
A typical agent definition contains:
- Persona — who the AI is (e.g., "a senior security engineer")
- Capabilities — what the agent can do
- Workflow — step-by-step process the agent follows
- Rules — guardrails and constraints
- Output format — how responses should be structured
- Tool references — MCP servers and skills the agent can use
AI agents vs prompts
A prompt is a one-off instruction ("review this code for bugs"). An agent is a persistent instruction set that defines a complete persona with capabilities, rules, and workflows. Agents are reusable across conversations and projects, while prompts are single-use.
Where can I find AI agents?
Agent Shelf is an open registry where developers publish, discover, and download AI agent definitions. Agents are available for coding, marketing, writing, DevOps, data analysis, security, and more.