use-cases/learn-a-new-concept.md +240 −0 added
1# Learn a new concept | Codex use cases
2
3[← All use cases](https://developers.openai.com/codex/use-cases)
4
5Copy page [Export as PDF](https://developers.openai.com/codex/use-cases/learn-a-new-concept/?export=pdf)
6
7Use Codex to study material such as research papers or courses, split the reading across subagents, gather context, and produce a Markdown report with diagrams.
8
9Intermediate
10
1130m
12
13Related links
14
15[Subagents](https://developers.openai.com/codex/subagents) [Subagent concepts](https://developers.openai.com/codex/concepts/subagents)
16
17## Best for
18
19 - Individuals learning about an unfamiliar concept
20- Dense source material that benefits from parallel reading, context gathering, diagrams, and a written synthesis
21- Turning a one-off reading session into a reusable Markdown report with citations, glossary terms
22
23## Skills & Plugins
24
25- [ImageGen](https://github.com/openai/skills/tree/main/skills/.curated/imagegen)
26
27 Generate illustrative, non-exact visual assets when a Markdown-native diagram is not enough.
28
29| Skill | Why use it |
30| --- | --- |
31| [ImageGen](https://github.com/openai/skills/tree/main/skills/.curated/imagegen) | Generate illustrative, non-exact visual assets when a Markdown-native diagram is not enough. |
32
33## Starter prompt
34
35 I want to learn a new concept from this research paper: [paper path or URL].
36 Please run this as a subagent workflow:
37- Spawn one subagent to map the paper's problem statement, contribution, method, experiments, and limitations.
38- Spawn one subagent to gather prerequisite context and explain the background terms I need.
39- Spawn one subagent to inspect the figures, tables, notation, and any claims that need careful verification.
40- Wait for all subagents, reconcile disagreements, and avoid overclaiming beyond the source material.
41 Final output:
42 - create `notes/[concept-name]-report.md`
43- include an executive summary, glossary, paper walkthrough, concept map, method diagram, evidence table, caveats, and open questions
44 - use Markdown-native Mermaid diagrams where diagrams help
45- use imagegen to generate illustrative, non-exact visual assets when a Markdown-native diagram is not enough
46 - cite paper sections, pages, figures, or tables whenever possible
47 Constraints:
48 - do not treat the paper as ground truth if the evidence is weak
49 - separate what the paper claims from your interpretation
50 - call out missing background, assumptions, and follow-up reading
51
52 I want to learn a new concept from this research paper: [paper path or URL].
53 Please run this as a subagent workflow:
54- Spawn one subagent to map the paper's problem statement, contribution, method, experiments, and limitations.
55- Spawn one subagent to gather prerequisite context and explain the background terms I need.
56- Spawn one subagent to inspect the figures, tables, notation, and any claims that need careful verification.
57- Wait for all subagents, reconcile disagreements, and avoid overclaiming beyond the source material.
58 Final output:
59 - create `notes/[concept-name]-report.md`
60- include an executive summary, glossary, paper walkthrough, concept map, method diagram, evidence table, caveats, and open questions
61 - use Markdown-native Mermaid diagrams where diagrams help
62- use imagegen to generate illustrative, non-exact visual assets when a Markdown-native diagram is not enough
63 - cite paper sections, pages, figures, or tables whenever possible
64 Constraints:
65 - do not treat the paper as ground truth if the evidence is weak
66 - separate what the paper claims from your interpretation
67 - call out missing background, assumptions, and follow-up reading
68
69## Introduction
70
71Learning a new concept from a dense paper or course requires more than just summarization. The goal is to build a working mental model: what problem it addresses, what the method actually does, which evidence supports it, what assumptions it depends on, and which parts you still need to investigate.
72
73Codex is useful here because it can automate the context gathering, and can turn complicated concepts into helpful diagrams or illustrations. This use case is also a good fit for [subagents](https://developers.openai.com/codex/concepts/subagents): one thread can read the paper for structure, another can gather prerequisite context, another can inspect figures and notation, and the main thread can reconcile the results into a report you can review later.
74
75For this use case, the final artifact should be something you can easily review: a Markdown file such as `notes/concept-report.md`, or a document of another format. It should include a summary, glossary, walkthrough, diagrams, evidence table, limitations, and open questions instead of ending with a transient chat answer.
76
77## Define the learning goal
78
79Start by naming the concept and the output you want. A narrow question makes the report more useful than a broad summary.
80
81For example:
82
83> I want to understand the main idea in this research paper, how the method works, why the experiments support or do not support the claim, and what I should read next.
84
85That scope gives Codex a concrete job. It should teach you the concept, but it should also preserve uncertainty, cite where claims came from, and separate the paper's claims from its own interpretation.
86
87## Running example: research paper analysis
88
89Suppose you want to learn about a paper about an unfamiliar model architecture. You want a report that lets you understand the concept at a glance, without having to read the whole paper.
90
91A good result might look like this:
92
93- `notes/paper-report.md` with the main explanation.
94- `notes/figures/method-flow.mmd` or an inline Mermaid diagram for the method.
95- `notes/figures/concept-map.mmd` or a small SVG that shows how the prerequisite ideas relate.
96- An evidence table that maps claims to paper sections, pages, figures, or tables.
97- A list of follow-up readings and unresolved questions.
98
99The point is to make the learning process more systematic and to leave behind a durable artifact.
100
101## Split the work across subagents
102
103Subagents work best when each one has a bounded job and a clear return format. Ask Codex to spawn them explicitly; Codex does not need to use subagents for every reading task, but parallel exploration helps when the paper is long or conceptually dense.
104
105For a research paper, a practical split is:
106
107- **Paper map:** Extract the problem statement, contribution, method, experiments, limitations, and claimed results.
108- **Prerequisite context:** Explain background terms, related concepts, and any prior work the paper assumes.
109- **Notation and figures:** Walk through equations, algorithms, diagrams, figures, and tables.
110- **Skeptical reviewer:** Check whether the evidence supports the claims, list caveats, and identify missing baselines or unclear assumptions.
111
112The main agent should wait for those subagents, compare their answers, and resolve contradictions. Codex will then synthesize the results into a coherent report.
113
114## Gather additional context deliberately
115
116When the paper assumes background you do not have, ask Codex to gather context from approved sources. That might mean local notes, a bibliography folder, linked papers, web search if enabled, or a connected knowledge base.
117
118If you're learning about an internal concept, you can connect multiple sources with [plugins](https://developers.openai.com/codex/plugins) to create a knowledge base.
119
120Keep this step bounded. Tell Codex what counts as a reliable source and what the final report should do with external context:
121
122- Define prerequisite terms in a glossary.
123- Add a short "background you need first" section.
124- Link follow-up readings separately from the paper's own claims.
125- Mark claims that come from outside the paper.
126
127## Generate diagrams for the report
128
129Diagrams are often the fastest way to check whether you really understand a concept. For a Markdown report, ask Codex for diagrams that stay close to the source material and are easy to revise.
130
131Good defaults include:
132
133- A concept map that shows prerequisite ideas and how they connect.
134- A method flow diagram that traces inputs, transformations, model components, and outputs.
135- An experiment map that connects datasets, metrics, baselines, and reported claims.
136- A limitations diagram that separates assumptions, failure modes, and open questions.
137
138For Markdown-first reports, ask for Mermaid when the destination supports it, or a small checked-in SVG/PNG asset when it does not. Ask Codex to use imagegen only when you need an illustrative, non-exact visual or something that doesn’t fit in a Markdown-native diagram.
139
140## Write the Markdown report
141
142Ask Codex to make the report self-contained enough that you can return to it later. A useful structure is:
143
1441. Executive summary.
1452. What to know before reading.
1463. Key terms and notation.
1474. Paper walkthrough.
1485. Method diagram.
1496. Evidence table.
1507. What the paper does not prove.
1518. Open questions and follow-up reading.
152
153The report should include source references wherever possible. For a PDF, ask for page, section, figure, or table references. If Codex cannot extract exact page references, it should say that and use section or heading references instead.
154
155## Use the report as a study loop
156
157The first report is a starting point. After reading it, ask follow-up questions and have Codex revise the artifact.
158
159Useful follow-ups include:
160
161- Which part of this method should I understand first?
162- What is the simplest toy example that demonstrates the core idea?
163- Which figure is doing the most work in the paper's argument?
164- Which claim is weakest or least supported?
165- What should I read next if I want to implement this?
166
167When the concept requires experimentation, ask Codex to add a small notebook or script that recreates a toy version of the idea. Keep that scratch work linked from the Markdown report so the explanation and the experiment stay together.
168
169Example prompt:
170
171Generate a script that reproduces a simple example from this paper.
172The script should be self-contained and runnable with minimal dependencies.
173There should be a clear output I can review, such as a csv, plot, or other artifact.
174If there are code examples in the paper, use them as reference to write the script.
175
176## Skills to consider
177
178Use skills only when they match the artifact you want:
179
180- `$jupyter-notebook` for toy examples, charts, or lightweight reproductions that should be runnable.
181- `$imagegen` for illustrative visual assets that do not need to be exact technical diagrams.
182- `$slides` when you want to turn the report into a presentation after the learning pass is done.
183
184For most paper-analysis reports, Markdown-native diagrams or simple SVG files are better defaults than a generated bitmap. They are easier to diff, review, and update when your understanding changes.
185
186## Suggested prompts
187
188**Create the Report Outline First**
189
190Before writing the full report, inspect [paper path] and propose the report outline.
191Include:
192- the core concept the paper is trying to explain
193- which sections or figures are most important
194- which background terms need definitions
195- which diagrams would help
196- which subagent tasks you would spawn before drafting
197Stop after the outline and wait for confirmation before creating files.
198
199**Build Diagrams for the Concept**
200
201Read `notes/[concept-name]-report.md` and add diagrams that make the concept easier to understand.
202Use Markdown-native Mermaid diagrams when possible. If the report destination cannot render Mermaid, create small checked-in SVG files instead and link them from the report.
203Add:
204- one concept map for prerequisites and related ideas
205- one method flow diagram for inputs, transformations, and outputs
206- one evidence map connecting claims to paper figures, tables, or sections
207Keep the diagrams faithful to the report. Do not add unverified claims.
208
209**Turn the Report Into a Study Plan**
210
211Use `notes/[concept-name]-report.md` to create a study plan for the next two reading sessions.
212Include:
213- what I should understand first
214- which paper sections to reread
215- which equations, figures, or tables need extra attention
216- one toy example or notebook idea if experimentation would help
217- follow-up readings and questions to resolve
218Update the report with a short "Next study loop" section.
219
220## Related use cases
221
222[
223
224### Coordinate new-hire onboarding
225
226Use Codex to gather approved new-hire context, stage tracker updates, draft team-by-team...
227
228Integrations Data](https://developers.openai.com/codex/use-cases/new-hire-onboarding)[
229
230### Generate slide decks
231
232Use Codex to update existing presentations or build new decks by editing slides directly...
233
234Data Integrations](https://developers.openai.com/codex/use-cases/generate-slide-decks)[
235
236### Analyze datasets and ship reports
237
238Use Codex to clean data, join sources, explore hypotheses, model results, and package the...
239
240Data Analysis](https://developers.openai.com/codex/use-cases/datasets-and-reports)