AI More detail C · Industry convention Platform: AI

Agent Loop

An Agent Loop is a controlled cycle. The agent checks state, acts with an allowed tool, reads the result, then stops or continues.

See how it works
You might call it agentic loopobserve-act looptool loop

See how it works

Original worked exampleAgent Loop

Another example

A planner reads schema data, proposes one safe check, and records the result. It stops at its step limit without touching production.

Main parts

  1. 01State and goal
  2. 02Bounded action
  3. 03Observation and stop

Use it when

Use it when each new result changes the next action. Set rules for stopping, cost, permission, and failure.

Do not use it when

Avoid endless retries, vague success, unchecked tool effects, and work after authority or evidence runs out.

Name used in code

state → decide → tool → observe → stop/continue

Before you ship

Check the model and prompt versions, source data, versioned evaluation set, and measures. Verify links to evidence, tool permissions, privacy, common failures, decline-to-answer and fallback behavior, human review, monitoring, cost, speed, and rollback.

Request you can copy

Outcome: Use or evaluate Agent Loop to make AI behavior measurable and tied to evidence. User context: A planner reads schema data, proposes one safe check, and records the result. It stops at its step limit without touching production. AI method or concept: Agent Loop. Why it fits: Use it when each new result changes the next action. Set rules for stopping, cost, permission, and failure. Do not use it when: Avoid endless retries, vague success, unchecked tool effects, and work after authority or evidence runs out. AI requirements: Define the input and source evidence. Set model and tool permissions. Use versioned evaluation data and measures. Define failure, decline-to-answer, privacy, speed, and cost limits. Operational safeguards: Keep a clear trace and hide sensitive log data. Show users a safe fallback. Mark steps that need human review. Define how to roll back the model or prompt. Acceptance criteria: Record baseline and target measures on a versioned evaluation set. Test edge cases and hostile inputs. Verify fallback, monitoring, permissions, and rollback. Evidence and limits (evidence boundary): No single official standard controls the whole term. Its meaning may change across teams, platforms, or frameworks. Unknowns to confirm: Target task, model and version, evaluation owner, source data, risk limit, tool permissions, and production fallback.

Check this request

Review the current use of Agent Loop. Definition: An Agent Loop is a controlled cycle. The agent checks state, acts with an allowed tool, reads the result, then stops or continues. Release checks: Check the model and prompt versions, source data, versioned evaluation set, and measures. Verify links to evidence, tool permissions, privacy, common failures, decline-to-answer and fallback behavior, human review, monitoring, cost, speed, and rollback. Before changing code, report the evidence you found, gaps, severity, and the smallest safe fix.

C
How official is this term?

Industry convention

People often use this term at work. The listed sources may support practice or limitations. They do not define every part of the term.

No single official standard controls the whole term. Its meaning may change across teams, platforms, or frameworks.

Scope
AI-agent engineering convention
Document status
stable
Checked on
2026-07-30

Evidence sources & scope

Authority source · ReAct authors · stable ReAct: Synergizing Reasoning and Acting in Language Models Scope: AI agent research Role here: Helpful background, not a definition Source covers: canonical name, definition, historical origin

Copy it yourself

The browser could not copy this. Select the request below and copy it yourself.