# Lilian Weng's 2023 blueprint for an agent, and the file each part leaves behind

**In June 2023, while at OpenAI, Lilian Weng published a long survey on her blog, Lil'Log, describing an LLM-powered agent as a brain (the model) plus three components: planning, memory and tool use. It's a widely cited early framework for what an agent is made of, and it's candid about where that framework still breaks.**

## What the post argues

Weng's overview sets the frame for the whole piece:

> “In a LLM-powered autonomous agent system, LLM functions as the agent’s brain, complemented by several key components: Planning ... Memory ... Tool use”.

Planning, in her account, covers both breaking a task into subgoals and reflecting on past actions to improve future ones. Memory splits into short-term (context the model can currently see, which she calls in-context) and long-term (usually stored outside the model, in a separate searchable database she calls a vector store). Tool use lets the model call out for anything a frozen set of weights can't provide on its own — current information, code execution, other APIs. Near the end, in a section she titles “Challenges,” she names a limitation squarely:

> “LLMs struggle to adjust plans when faced with unexpected errors, making them less robust compared to humans who learn from trial and error.”

Elsewhere, in a tool-use case study about a chemistry agent called ChemCrow, she flags a narrower problem: an LLM-based evaluation rated it about equal to GPT-4, while human experts judged ChemCrow far better on correctness. Her conclusion is about self-evaluation, not about her “reflection” component specifically:

> “The lack of expertise may cause LLMs not knowing its flaws and thus cannot well judge the correctness of task results.”

Weng doesn't mention MarsDawn anywhere in this post, and doesn't recommend any Markdown tool.

## Our reading, not Weng's

Weng is describing agent architecture in 2023, not writing about anyone reading agent output — she doesn't mention a person checking a file at all. But her own three components map onto three different things you might find yourself reading. Planning tends to leave you a document to read before it runs — the plan itself, sometimes with a “reflection” or self-review already folded in. Memory is usually invisible unless the agent keeps a running scratch file as its long-term store, in which case that file is worth opening on its own, since it can carry forward an old, wrong assumption across many later steps without saying so. Tool use tends to leave you a report of what ran and what came back — closer to a transcript than a plan.

Her point about plans not adjusting to unexpected errors is, read from your side, a reason a plan you approved yesterday can be stale today: if something the plan didn't anticipate happened between then and now, the agent may keep going anyway rather than replanning, and the report at the end can describe the original plan's success without describing the detour. That's our inference, not a claim she makes — she's writing about a model's own robustness, not about what a reader should watch for.

## Where MarsDawn helps, and where it doesn't

MarsDawn has no AI model inside, so it can't tell you whether a plan has quietly drifted from what actually happened, and it doesn't distinguish a planning file from a memory file from a tool-use report — that's a read on the content, which is yours to make. What it does: the Outline tab (View ▸ Show Sidebar, ⌃⌘S) shows a long plan's shape at a glance, the source and rendered preview sit side by side (⌘2) with Mermaid and KaTeX drawn out, and if the agent rewrites the file mid-read, MarsDawn reloads it and keeps your place, as long as you have no unsaved edits of your own — useful specifically because a plan that's been silently revised is exactly the failure mode her “Challenges” section describes from the model's side.

## Try it

MarsDawn is on the Mac App Store. The free `marsdawn` command-line tool works today:

```
brew install redtear1115/tap/marsdawn
```

It exports Markdown to PDF without the app.

[Command Line](/cli/) · Know before you buy: [What MarsDawn doesn't do](/limits/)

## Next

- Which documents different agent design patterns tend to hand you: [Four agent design patterns and the documents each one hands you](/agent-design-patterns/)
- The five-minute way to review a plan before it runs: [Reviewing an agent plan in five minutes](/reviewing-agent-plans/)
- Back to the series: [Editor's reading notes](/reading-notes/)

## Sources

- Lilian Weng, “LLM Powered Autonomous Agents,” Lil'Log, June 23, 2023: [https://lilianweng.github.io/posts/2023-06-23-agent/](https://lilianweng.github.io/posts/2023-06-23-agent/) (fetched and quoted 2026-09-26; she was at OpenAI when she wrote it, described here only as she was then).

## More

- [MarsDawn](https://marsdawn.southern-light.dev/index.md): Read and review the Markdown your agent wrote, natively on Mac: live split preview, Mermaid, KaTeX, Quick Look, PDF export. Free to try, USD 4.99 once.
- [Features](https://marsdawn.southern-light.dev/features/index.md): A native Mac Markdown editor: live preview, Mermaid and math, themes, plus clean PDF and print pages for meetings, margin notes and paper records.
- [Your writing stays on your Mac](https://marsdawn.southern-light.dev/yours/index.md): MarsDawn has no account, no sync and no cloud. Your Markdown documents stay on your Mac, in the files and folders you choose.
- [Try free, pay once](https://marsdawn.southern-light.dev/pay-once/index.md): MarsDawn is free to download. Try everything for 14 days, then unlock it once for USD 4.99. No subscription, no account.
- [PDF export](https://marsdawn.southern-light.dev/pdf/index.md): Export Markdown as a PDF or print it on your Mac, with Mermaid diagrams and highlighted code. Page breaks avoid splitting short code blocks.
- [A Mac app](https://marsdawn.southern-light.dev/native/index.md): A Markdown editor that is a real Mac app: native windows and tabs, autosave, version history, Quick Look in Finder and a text editor that behaves like a Mac.
- [What MarsDawn doesn't do](https://marsdawn.southern-light.dev/limits/index.md): No sync, no editing on iPhone or iPad, no plugins, no accounts. Themes, but no custom CSS. Know before you buy.
- [Support](https://marsdawn.southern-light.dev/support/index.md): Get help with MarsDawn, the Markdown editor for macOS.
- [Privacy Policy](https://marsdawn.southern-light.dev/privacy/index.md): MarsDawn does not collect personal data. Your documents and settings stay on your Mac.
- [How to open .md files on Mac](https://marsdawn.southern-light.dev/view-markdown-on-mac/index.md): How to open a .md file on a Mac: double-click, set a default app, preview in Finder. For .md files an AI agent wrote, read them rendered in MarsDawn.
- [Mermaid preview on Mac](https://marsdawn.southern-light.dev/mermaid-preview-mac/index.md): See Mermaid diagrams drawn next to your Markdown on a Mac: live preview, Quick Look and PDF export. A broken diagram shows its error and the line.
- [Quick Look for Markdown](https://marsdawn.southern-light.dev/quicklook/index.md): MarsDawn's Quick Look renders Markdown in Finder when you press Space, with Mermaid, KaTeX and highlighted code. It keeps working after the trial ends.
- [Markdown to PDF](https://marsdawn.southern-light.dev/markdown-to-pdf/index.md): Convert Markdown to PDF on a Mac with the free marsdawn command-line tool. Install it with Homebrew and run one command: tables, math, Mermaid and code.
- [MarsDawn compared: VS Code preview, Obsidian, QLMarkdown, LynxOffice](https://marsdawn.southern-light.dev/vs/index.md): For anyone choosing how to read and review Markdown on a Mac, often what an agent wrote. Every cell below comes from the vendor's own pages or our own check, all on 16 October 2026; products change, so a cell can go stale. If you spot a mistake, write to the address on Support.
- [MacMD Viewer vs. MarsDawn](https://marsdawn.southern-light.dev/vs/macmd-viewer/index.md): MacMD Viewer is a read-only Markdown viewer for USD 19.99. MarsDawn edits and previews side by side: free for 14 days, then USD 4.99 once.
- [Best Markdown viewer for Mac (2026)](https://marsdawn.southern-light.dev/best-markdown-viewer-mac/index.md): Eight Markdown viewers for Mac compared by price, Mermaid, math, PDF, live reload, Quick Look and native vs Electron, with who each one suits.
- [Command Line](https://marsdawn.southern-light.dev/cli/index.md): The free marsdawn command-line tool for Mac: export Markdown to PDF from a shell, a script or an LLM agent, with JSON output. Install it with Homebrew.
- [marsdawn for agents](https://marsdawn.southern-light.dev/cli/agents/index.md): A reference for AI agents and scripts that call marsdawn to turn Markdown into PDF: commands, JSON output, schemas, exit codes and requirements.
- [Agent skill](https://marsdawn.southern-light.dev/cli/skill/index.md): One file your coding agent loads to open Markdown it wrote in MarsDawn for your review, and to install marsdawn, export Markdown to PDF and read the JSON result.
- [MCP server](https://marsdawn.southern-light.dev/cli/mcp/index.md): marsdawn has no AI model of its own, so it doesn't matter which agent wrote the Markdown. Call it from the CLI, a skill file, or the marsdawn-mcp MCP server: all three run the same export.
- [Token-efficient review](https://marsdawn.southern-light.dev/token-efficient-review/index.md): A person reviews the rendered page in MarsDawn, never read back into the agent's context. The tool call itself returns a compact JSON result, not the rendered content, so calling it is cheap too.
- [Viewing Markdown elsewhere vs. MarsDawn](https://marsdawn.southern-light.dev/vs/markdown-preview-tools/index.md): How MarsDawn compares to reading Markdown in VS Code's built-in preview, a browser extension, or Claude Desktop's file preview: what each renders, and what it takes to open one file.
- [Themes](https://marsdawn.southern-light.dev/themes/index.md): Four built-in preview themes, each light and dark, plus a curated gallery you open from Settings in MarsDawn for Mac. Quick Look uses your theme.
- [Build a theme](https://marsdawn.southern-light.dev/themes/new/index.md): Pick colours and a handful of style options, see them applied to a sample document live, and submit your theme as a GitHub issue. No install, no git.
- [Make a theme with your agent](https://marsdawn.southern-light.dev/themes/agents/index.md): MarsDawn theme reference for AI agents: each theme.json field with type and range, the CLI loop that checks and renders it, a hand-off link, and how to submit.
- [Theme gallery](https://marsdawn.southern-light.dev/themes/gallery/index.md): Browse preview themes for MarsDawn, each reviewed by the developer, filter them by scenario, and report a problem with one. Build your own in the browser.
- [Sharing exported PDFs](https://marsdawn.southern-light.dev/sharing-exported-pdfs/index.md): Export an agent's Markdown to PDF and hand it to a colleague who doesn't read Markdown and won't install anything. No syntax, no app and no account needed to open it.
- [Why AI output still needs a human reader](https://marsdawn.southern-light.dev/reviewing-ai-output/index.md): AI-written Markdown still has to be understood by a person, not trusted on sight. MarsDawn pairs the rendered page with the source, and draws Mermaid diagrams and KaTeX math, so structure is legible at a glance.
- [Reading what your agent hands back](https://marsdawn.southern-light.dev/reading-agent-output/index.md): AI agents hand back their work as Markdown: plans, specs, progress reports. What people who build agents say about checkpoints and failures, why that output is hard to read, and a five-minute checklist for reviewing a plan.
- [Agent transparency](https://marsdawn.southern-light.dev/agent-transparency/index.md): Anthropic's guide to building agents asks for transparency: show the planning steps. What it says, what it doesn't, and why the steps usually end up as a Markdown file someone has to read.
- [Reviewing an agent plan](https://marsdawn.southern-light.dev/reviewing-agent-plans/index.md): A six-step way to review the plan an AI agent hands you before it runs, in about five minutes and in any editor, with a worked example.
- [Agent design patterns](https://marsdawn.southern-light.dev/agent-design-patterns/index.md): Reflection, tool use, planning and multi-agent collaboration, as Andrew Ng described them, and what each tends to hand back for you to read.
- [Changelog](https://marsdawn.southern-light.dev/changelog/index.md): What changed in each version of the MarsDawn app and the free marsdawn CLI.
- [Roadmap](https://marsdawn.southern-light.dev/roadmap/index.md): What's in progress and what's next for MarsDawn on Mac, iPhone and iPad, by version. No dates, and unscheduled ideas are marked.
- [Editor's reading notes](https://marsdawn.southern-light.dev/reading-notes/index.md): Six short notes on what the people building AI agents actually argue — Anthropic, Chip Huyen, Lilian Weng, Harrison Chase, LangChain and Andrew Ng — and what each one means for the person who has to read what such an agent hands back.
- [Reading notes: Anthropic](https://marsdawn.southern-light.dev/reading-notes/anthropic-building-effective-agents/index.md): Anthropic's December 2024 guide for people building agents separates workflows from agents and describes five workflow patterns, including one where a second LLM call reviews the first. What that means for what lands in your folder.
- [Reading notes: Chip Huyen](https://marsdawn.southern-light.dev/reading-notes/chip-huyen-agents/index.md): Chip Huyen's January 2025 essay on agents splits their actions into read-only and write actions. Why that split is a fast way to spot the line in a plan worth a closer look before you approve it.
- [Reading notes: Harrison Chase](https://marsdawn.southern-light.dev/reading-notes/harrison-chase-what-is-an-agent/index.md): Harrison Chase's 2024 definition of an agent and his spectrum of agentic behavior, and his case for observability as a system moves along it — read from the side of whoever reads the file it hands back.
- [Reading notes: LangChain (Jess Ou)](https://marsdawn.southern-light.dev/reading-notes/langchain-what-is-an-agent/index.md): LangChain's 2026 “What is an AI agent?” by Jess Ou echoes Harrison Chase's 2024 definition and describes a pipeline for evaluating agents automatically. Where that pipeline still hands a step to a person, and where it doesn't.
- [Reading notes: Andrew Ng](https://marsdawn.southern-light.dev/reading-notes/andrew-ng-design-patterns/index.md): Across five letters in The Batch, Andrew Ng ranks reflection, tool use, planning and multi-agent collaboration by how reliable and predictable he finds each one — and what that ranking suggests about how closely to check each one's output.
- [Templates](https://marsdawn.southern-light.dev/templates/index.md): Markdown templates for the documents an agent writes and you read: a spec, a flowchart and meeting notes, each with a prompt for your agent.
- [Spec template](https://marsdawn.southern-light.dev/templates/spec/index.md): A Markdown spec template with requirements, a Mermaid flow diagram and acceptance criteria. Your agent fills it in; you review it in MarsDawn.
- [Flowchart template](https://marsdawn.southern-light.dev/templates/flowchart/index.md): A Mermaid flowchart template in Markdown, with the steps written out below it. Preview it on a Mac and export it to PDF.
- [Meeting notes template](https://marsdawn.southern-light.dev/templates/meeting-notes/index.md): A Markdown meeting notes template with decisions and action items, each with an owner. Your agent writes it up; you check it in MarsDawn.
- [繁體中文](https://marsdawn.southern-light.dev/zh-hant/reading-notes/lilian-weng-llm-agents/index.md): Lilian Weng 2023 年被廣泛引用的整理，把 LLM agent 描述成大腦加上規劃、記憶、工具使用。每個部件通常會留給你讀什麼，以及她點名的一個限制：計畫遇到意外不太會調整。
- [简体中文](https://marsdawn.southern-light.dev/zh-hans/reading-notes/lilian-weng-llm-agents/index.md): Lilian Weng 2023 年被广泛引用的整理，把 LLM 代理描述成大脑加上规划、记忆、工具使用。每个部件通常会留给你读什么，以及她点名的一个限制：计划遇到意外不太会调整。
- [日本語](https://marsdawn.southern-light.dev/ja/reading-notes/lilian-weng-llm-agents/index.md): Lilian Weng が 2023 年に書いた、広く引用されているサーベイは、LLM エージェントを、脳とプランニング、記憶、ツール利用の組み合わせとして描く。各部分が普通あなたに何を読ませることになるか、そして計画が予想外の事態に調整できないという、彼女自身が挙げる限界。
- [Deutsch](https://marsdawn.southern-light.dev/de/reading-notes/lilian-weng-llm-agents/index.md): Lilian Wengs viel zitierte Umfrage von 2023 beschreibt einen LLM-Agenten als Gehirn plus Planung, Gedächtnis und Werkzeugnutzung. Was jeder Teil dir typischerweise zum Lesen hinterlässt, und die Grenze, die sie bei Plänen nennt, die sich Überraschungen nicht anpassen.
- [Français](https://marsdawn.southern-light.dev/fr/reading-notes/lilian-weng-llm-agents/index.md): L’enquête très citée de Lilian Weng en 2023 décrit un agent LLM comme un cerveau plus planification, mémoire et usage d’outils. Ce que chaque partie tend à vous laisser à lire, et la limite qu’elle nomme dans les plans qui ne s’ajustent pas aux surprises.
- [Español](https://marsdawn.southern-light.dev/es/reading-notes/lilian-weng-llm-agents/index.md): La encuesta muy citada de Lilian Weng de 2023 describe un agente LLM como un cerebro más planificación, memoria y uso de herramientas. Qué suele dejarte cada parte para leer, y el límite que nombra en planes que no se ajustan a las sorpresas.
- [한국어](https://marsdawn.southern-light.dev/ko/reading-notes/lilian-weng-llm-agents/index.md): Lilian Weng의 널리 인용되는 2023년 서베이는 LLM 에이전트를 뇌와 계획·기억·도구 사용으로 그립니다. 각 부분이 당신에게 남기기 쉬운 파일, 그리고 예상치 못한 일에 조정되지 않는 계획에서 그녀가 짚는 한계.