paper-read
Skill182 starsChanged 41 days agoMITView on GitHub
Summarize and critically analyze academic papers supplied as PDFs, text, arXiv links, or journal URLs. Ground claims in the inspected version and source locations; support focused questions, deeper appraisal, and cross-paper comparison.
4 other projects use a version of this.
Adopted by
Most recently taken up by
- DaeSeokSong/oh-my-ouroborostheir own version, first seen Nov 1, 2026
- genliusrocks/paper_gardentheir own version, first seen Oct 15, 2026
- ZeyuLing/research_autocode_skillstheir own version, first seen Jun 28, 2026
- andrehuang/researcher-packtheir own version, first seen Apr 22, 2026
Other versions4 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 4
- andrehuang/researcher-pack★ 50same name
- ZeyuLing/research_autocode_skills★ 9same name
- DaeSeokSong/oh-my-ouroboros★ 3same name
- genliusrocks/paper_garden★ 1same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository4 other files it gives its agents.
- latex-fmtSkill
- latex-polishSkill
- latex-rescueSkill
- pdf2texSkill
The file
We split this skill into the rules it gives an agent
6 rules. About 0 are common, 4 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 23
Use the supplied text or accessible PDF.
- Distinctive: line 32
If extraction loses columns, math, tables, or figures, inspect rendered pages and state what remains unreadable.
- Very specific: line 42
Read reading-framework.md for mode-specific strategies and article-type adjustments.
- Distinctive: line 57
Treat publication status as context, not a substitute for evidence.
- Distinctive: line 61
Verify external comparisons through inspected primary sources.
- Very specific: line 77
Use latex-polish or latex-fmt only for an additional requested manuscript task and only when the corresponding skill is available.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Choose the depth
Match the user's question; a focused question need not produce a full-paper report.
| Mode | Read for | Output |
skim | Problem, approach, main results, figures, stated limitations | Concise overview with coverage limits. |
read (default) | Method, setup, results, and claim/evidence alignment | Structured account with source locations. |
deep | Assumptions, equations, ablations, implementation details, supplements | Critical appraisal and unresolved questions. |
Depth describes coverage, not a guaranteed duration. A skim cannot certify methodological soundness. Use the user's language unless they request otherwise.
Acquire and identify the evidence
Use the supplied text or accessible PDF. For a link, obtain the paper through
available browsing/download tools; record title, authors, identifier, exact
version, publication status, and available main/supplementary files.
An arXiv vN suffix identifies a specific version; preserve it instead of silently
switching to the newest paper. New and legacy identifiers are described in the
official arXiv identifier guide.
Extract with an available PDF reader, PyMuPDF, or Poppler. Use task-local files
and paths appropriate to the host; do not assume /tmp or a particular shell.
If extraction loses columns, math, tables, or figures, inspect rendered pages
and state what remains unreadable. A missing tool does not require installing
software when supplied text or another reader is sufficient.
For inaccessible/paywalled material, look for a legitimate author or repository copy, identify version differences, or request the relevant text. Do not infer the full paper from an abstract or pretend the supplement was inspected.
Analyze the inspected material
Read reading-framework.md for mode-specific
strategies and article-type adjustments. For read and deep, use the relevant
items from critical-appraisal.md.
Connect the problem, method, assumptions, training/inference setup, data, baselines, metrics, and results. Preserve units, denominators, uncertainty, dataset splits, and relative versus absolute improvements. Do not call a result statistically significant from its size alone.
For important claims, provide the claim, supporting section/table/figure/equation, what the evidence establishes, and any gap. Use printed page numbers where available and distinguish them from PDF page indexes. Distinguish an author's claim, your interpretation, an algebraic consistency check, and reproduced evidence. Reading a paper is not an independent reproduction or proof audit.
Treat publication status as context, not a substitute for evidence. The same criteria apply to a preprint and a peer-reviewed article. A repository listing or arXiv upload alone does not establish peer-review status.
Verify external comparisons through inspected primary sources. If only the paper's related-work account is available, attribute that comparison to the authors; do not invent a literature history or recommend unread papers as evidence.
Compare papers and report
For multiple papers, align task, dataset/split, metric, evaluation protocol, resources, and version before comparing numbers. Label incomparable results. Explain whether apparently contradictory claims involve different settings.
Deliver an answer centered on the user's question, key conclusions with source locations, strengths and limitations supported by evidence, inspected coverage, and unresolved uncertainties. For implementation requests, separate published settings, public-code defaults, and missing details. Relate findings to the user's stated research goal without assuming one from the author list or venue.
Use latex-polish or latex-fmt only for an additional requested manuscript task
and only when the corresponding skill is available.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.