Paper Deep Reader skill
A skill for deep-reading one scientific paper: six sections, key figures in the text and each claim checked against the source.
- 116 ★ on the repository
- Last commit
- MIT
Minimal access
Why this level
- The connected model processes the paper's text and images.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/paper-deep-readerDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewMinimal accessThe tool runs nothing and makes no network calls.
- The connected model processes the paper's text and images.
This is third-party code. Review the repository files before installing.
What it does
Paper Deep Reader turns a single scientific paper into a coherent six-part report: one-line summary, background and motivation, method, experiments, contribution and conclusion. Key figures and tables sit inside the text, each followed by what it shows and how it supports the conclusion. The skill separates author claims, direct evidence, its own inferences and outside background. Modes are `brief`, `explain`, `audit` and `targeted`, with routes for models with and without vision. Reader settings live in `.paper-reader.yaml`.
Who it is for. Researchers and students who need to understand a paper outside their field and check what its conclusions rest on.
Good fit when
- You must quickly decide whether a paper is worth reading in full.
- A paper from another field needs terms and an example explained.
- A reviewer checks how a formula maps to a block in a figure.
- A report with full analysis of all key figures is needed.
Not a fit when
- You want a verbatim translation of the paper.
- You need a survey of several works, since the skill targets one.
- The PDF is unreadable and neither text nor images can be extracted.
Example request
Break this PDF down into a full report for a reader with no background and explain every key figure.Limitations
A model writes the report, so check formulas and numbers against the original. Without a vision model figures are analysed from captions and text, and the skill says so. Extracting images from a PDF needs Python and the dependencies in `requirements.txt`. The docs are in Chinese, and reports can be requested in any language. The skill does not replace reading the paper yourself, it prepares an analysis worth checking against the source.
How to disable. Delete the skill directory, for example `~/.codex/skills/paper-deep-reader`, and the `.paper-reader.yaml` file from the project.
FAQ
How do the modes differ?
`brief` gives the core idea on one screen, `explain` the full six-part report, `audit` adds analysis of all figures and a reproducibility check, and `targeted` answers one question.
Does the skill work without a vision model?
Yes, there is a text-only route. Figures are read from captions and descriptions and the report warns about it.
How do I update the skill?
Run `git -C ~/.codex/skills/paper-deep-reader pull --ff-only` with your actual install path.
What audience is the report written for?
By default a reader with no background but a doctoral technical ceiling. Settings change in `.paper-reader.yaml`.
What does the skill not do?
It does not translate the paper verbatim and does not write a survey of several works. It analyses one paper.
Discussion
No questions or reviews yet.
Related
SkillsA skill that deep-reads a single paper and builds a structured Obsidian-style research note
Reads one biomedical paper figure by figure: why each experiment was run, what the data support, what to borrow
A skill set for academic papers: deep method analysis, visual slides and a coherent write-up from a paper PDF
paper-reading-skill turns one arXiv AI or robotics paper into a report with each claim tied to evidence in the source.
The paper-spine skill takes a topic, materials and data through a scientific paper: literature, text, figures, review and Word, LaTeX and PDF files
A Claude Code skill for scientific literature reviews: an interview, Scopus and OpenAlex search, DOI checks and a final docx file
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| Paper Deep Reader (this card) | no | 116 | |
| DeepPaperNote | no | 1,362 | builds a structured Obsidian-style note; Paper Deep Reader writes a six-part report with figures and separated evidence |
| paper-reading-skill | no | 151 | a simpler paper walkthrough; here the brief, explain and audit modes and a route without a vision model |
README in short
The report moves from a clear problem to the method and evidence rather than following the paper's section order. Each conclusion states how strongly it is supported.