agent
An LLM-driven program that plans, calls tools, observes results, and loops until it reaches a goal.
Definition
An agent is a program built around an LLM that can plan multi-step actions, call external tools, observe results, and loop until a goal is met. Unlike a single-turn prompt, an agent maintains state across iterations and decides its own next step. Common agent stacks: Claude Code, LangChain agents, Anthropic's Agent SDK, Make.com AI scenarios.
When to use
Use an agent when the task can't be solved in a single LLM call — when the model needs to discover information, take actions, evaluate results, and adapt before finishing.
See also
- MCP — Anthropic's open standard for connecting LLMs to external tools and data sources over JSON-RPC.
- Claude Code — Anthropic's official terminal-based coding agent, with MCP support and a plugin ecosystem.
- LangChain — Python and JavaScript framework for building LLM applications — chains, agents, memory, retrieval, and tool use.
Mentioned in
- agency-os: the Notion board that plans your work, then ships it
- AI search invisibility: 7 reasons your site isn't cited (and how to fix each)
- Cursor rules not loading: 4 causes and how to fix each
- How to fix Claude Code SessionStart hooks firing without context
- How to fix CrewAI hierarchical DelegateWorkToolSchema validation error
- How to fix LangChain MultiServerMCPClient NotImplementedError on Windows
- How to fix MCP Token expired without refresh token in Claude Code
- How to fix UiPath Studio 2024.10.1 EEUI / 1603 install error
- How to set up an MCP server in Claude Code on Windows
- Launching Citation Intelligence MCP: free, self-hosted AI citation tracking
- Launching content-distribution-mcp: one finished post, eight channels
- Notion Custom Agents vs Claude Code: when to use each in 2026
- Three Claude skills for end-to-end Ghost blog publishing
- We introspected 922 MCP servers. Here's what we found.
- What is the Model Context Protocol (MCP)? A 2026 primer