Dictionary
Reflection
An agent critiquing and revising its own output before returning.
1 min readupdated 2026-07-04
/ quick answer
Reflection adds a second pass: the agent reads its draft, lists flaws, and rewrites. Costs 2x tokens; often worth it for writing, code, and analysis. An agent critiquing and revising its own output before returning.
An agent critiquing and revising its own output before returning. Reflection adds a second pass: the agent reads its draft, lists flaws, and rewrites. Costs 2x tokens; often worth it for writing, code, and analysis. In practice: Writer agent drafts an article, then reflects: 'weak intro, thin evidence in section 3', then rewrites. This dictionary node is part of the Onexial knowledge graph and links to related concepts, workflows and tools below.
Definition
Reflection adds a second pass: the agent reads its draft, lists flaws, and rewrites. Costs 2x tokens; often worth it for writing, code, and analysis.
Example
Writer agent drafts an article, then reflects: 'weak intro, thin evidence in section 3', then rewrites.
/ frequently asked
What is Reflection?
Reflection adds a second pass: the agent reads its draft, lists flaws, and rewrites. Costs 2x tokens; often worth it for writing, code, and analysis.
What is an example of Reflection?
Writer agent drafts an article, then reflects: 'weak intro, thin evidence in section 3', then rewrites.
Why does Reflection matter for AI and automation?
An agent critiquing and revising its own output before returning. It connects to the workflows, prompts and tool stacks linked on this page, so you can move from definition to execution without leaving Onexial.
↳ connected nodes
Comparison↳ linked
OpenAI vs Anthropic for Agents
Which provider builds better production agents in 2026.
Comparison↳ linked
CrewAI vs AutoGen vs LangGraph
Three approaches to multi-agent orchestration.
Dictionary↳ linked
MCP (Model Context Protocol)
Open protocol that lets LLMs connect to tools, data sources and apps through a standard interface.
Dictionary↳ linked
Agentic RAG
RAG where an agent decides what to retrieve, when, and from which source — instead of a single static query.
Dictionary↳ linked
Agentic Workflow
A workflow where an LLM decides the next step instead of a hard-coded path.
Dictionary↳ linked
Tool Use
An LLM invoking external functions to act beyond text generation.