Scaffolded agent

A scaffolded agent is built on top of a strong general-purpose framework (LangGraph, OpenAI Agents SDK) that provides the agent loop, tracing, and tool-use plumbing — letting developers focus on the domain logic.

Building agents from scratch in 2026 is rarely worth it. Scaffolded agents inherit the framework's agent loop, planner-executor split, checkpoint management, human-in-the-loop hooks, tracing, and tool-use plumbing — and the developer writes the domain-specific tools and prompts. The result: faster development, fewer reliability bugs, easier observability. Best practice is to pick the framework that matches your model (LangGraph for cross-model, OpenAI Agents SDK for OpenAI-native, Claude Agent SDK for Anthropic-native) and to add custom logic only where the framework can't reach.

When to use scaffolded agent

• Any production agent.
• Teams without dedicated agent infrastructure engineering.

Common mistakes

• Re-inventing the agent loop on top of raw chat APIs.
• Overriding framework behaviour without understanding what you lose (tracing, checkpoints, eval hooks).

FAQ

Related terms

• AI agent — An AI agent is a system where a language model autonomously plans and executes a sequence of tool calls to accomplish a goal.
• Agent loop — An agent loop is the repeating cycle of an AI agent — observe state, decide on an action (usually a tool call), execute, observe the result, and repeat — until a goal is reached or a stop condition fires.
• Tool use (LLM) — Tool use is the umbrella term for any LLM mechanism that lets the model invoke external functions, APIs, or services — function calling, code interpreter, MCP servers, browser actions.
• Evals (LLM evaluations) — Evals are systematic tests that measure how well a language model or LLM-powered system performs on a defined task using a golden set of inputs and reference outputs.

Last updated: 2026-06-01. Raw markdown: https://promtable.com/glossary/scaffolded-agent.md.