agentleFS
DeepXivDeepXiv/deepxiv_sdk/skills/deepxiv-trending-digest/SKILL.md

deepxiv-trending-digest

Skill843 starsChanged 8 months agoMITView on GitHub

Summarize recent hot academic papers using deepxiv trending, brief, head, and section reads, then produce a markdown digest highlighting what each paper is about and which papers deserve deeper reading.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

Show its rules and text

We split this skill into the rules it gives an agent

26 rules. About 0 are common, 13 are distinctive, and 13 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈13

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 13

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use this skill when the user wants a recent hot-paper roundup, trending paper summary, weekly paper digest, or a markdown report based on …

    line 8
  2. Very specific:

    Read only the most relevant sections with deepxiv paper <id> --section ...

    line 18
  3. Very specific:

    Write a clean .md summary with recommendations

    line 19
  4. Distinctive:

    Start with a small, recent set unless the user asks otherwise.

    line 25
  5. Very specific:

    Use --days 7 for "recent hot papers"

    line 32
  6. Very specific:

    Use --limit 10 for a manageable first pass

    line 33
  7. Very specific:

    If the user asks for a broader roundup, use --days 14 or --days 30

    line 34
  8. Very specific:

    Do not jump to full text yet.

    line 53
  9. Distinctive:

    After reading briefs, choose the top 1-3 papers for deeper inspection.

    line 57
  10. Distinctive:

    If a paper sounds incremental, unclear, or out of scope, keep it in the digest but skip deep reading.

    line 67
  11. Very specific:

    Use --head to decide whether to drill down and which section matters most.

    line 77
  12. Distinctive:

    Look for sections such as:

    line 79
  13. Distinctive:

    Prefer section reads over full paper reads.

    line 85
  14. Distinctive:

    Read at most 1-2 sections per paper unless the user explicitly asks for a deep dive.

    line 89
  15. Very specific:

    Read Introduction if the contribution is still fuzzy

    line 100
  16. Very specific:

    Read Method if the core idea matters

    line 101
  17. Very specific:

    Read Results if the claim sounds strong and needs verification

    line 102
  18. Very specific:

    Read Limitations or Discussion if tradeoffs matter

    line 103
  19. Distinctive:

    Write a markdown file in the current workspace unless the user gave a different path.

    line 109
  20. Distinctive:

    Keep the digest skimmable

    line 173
  21. Distinctive:

    Prefer short paragraphs and flat bullets

    line 174
  22. Distinctive:

    Do not pretend to have verified claims you have not checked

    line 178
  23. Distinctive:

    If briefs are enough for the user's request, stop there

    line 182
  24. Very specific:

    Use --head only for the most promising papers

    line 183
  25. Very specific:

    Use --section only after --head suggests a high-value section

    line 184
  26. Distinctive:

    Do not read full paper markdown unless the user explicitly asks for a full analysis

    line 185

≈ “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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

DeepXiv Trending Digest

Use this skill when the user wants a recent hot-paper roundup, trending paper summary, weekly paper digest, or a markdown report based on DeepXiv trending papers.

Goal

Turn recent DeepXiv trending papers into a concise markdown digest:

  1. Find currently hot papers with deepxiv trending
  2. Brief each candidate with deepxiv paper <id> --brief
  3. Select the most promising papers for deeper inspection
  4. Inspect structure with deepxiv paper <id> --head
  5. Read only the most relevant sections with deepxiv paper <id> --section ...
  6. Write a clean .md summary with recommendations

Default Workflow

1. Pull trending papers

Start with a small, recent set unless the user asks otherwise.

deepxiv trending --days 7 --limit 10 --json

Default heuristics:

  • Use --days 7 for "recent hot papers"
  • Use --limit 10 for a manageable first pass
  • If the user asks for a broader roundup, use --days 14 or --days 30

2. Brief every candidate

For each selected arXiv ID, fetch a brief first:

deepxiv paper <arxiv_id> --brief

Capture:

  • title
  • arXiv ID
  • publish date
  • keywords
  • TLDR
  • citations if present
  • GitHub URL if present

Do not jump to full text yet. --brief is the default screening step.

3. Rank for deeper reading

After reading briefs, choose the top 1-3 papers for deeper inspection.

Use signals like:

  • strong novelty or surprising result
  • practical relevance
  • likely user interest
  • unusually clear contribution
  • useful code release
  • especially strong social or research momentum

If a paper sounds incremental, unclear, or out of scope, keep it in the digest but skip deep reading.

4. Inspect paper structure

For the chosen papers:

deepxiv paper <arxiv_id> --head

Use --head to decide whether to drill down and which section matters most.

Look for sections such as:

  • Introduction
  • Method / Approach / Framework
  • Experiments / Results / Evaluation
  • Discussion / Limitations

Prefer section reads over full paper reads.

5. Read only high-value sections

Read at most 1-2 sections per paper unless the user explicitly asks for a deep dive.

Examples:

deepxiv paper <arxiv_id> --section Introduction
deepxiv paper <arxiv_id> --section Method
deepxiv paper <arxiv_id> --section Results

Selection guidance:

  • Read Introduction if the contribution is still fuzzy
  • Read Method if the core idea matters
  • Read Results if the claim sounds strong and needs verification
  • Read Limitations or Discussion if tradeoffs matter

Avoid reading everything.

Markdown Output

Write a markdown file in the current workspace unless the user gave a different path.

Recommended filename:

trending-paper-digest-YYYY-MM-DD.md

Recommended structure:

# Trending Paper Digest

Date: YYYY-MM-DD
Window: Last 7 days
Source: deepxiv trending

## Executive Summary

- 2-4 bullets with the main themes across the trending list
- Mention the 1-3 most promising papers

## Papers Reviewed

### 1. Paper Title (`arXiv:xxxx.xxxxx`)

What it is about:
Short paragraph based mainly on `--brief`.

Why it matters:
- Bullet
- Bullet

Worth deeper reading?
Yes/Maybe/No, with one sentence.

If deeper review was done:

Sections checked:
- Introduction
- Method

Deeper notes:
Short paragraph with the key insight, evidence, or caveat.

### 2. ...

## Recommended Deep Dives

### Paper Title

- Why it stands out
- Which section to read next
- What question it could answer

## Cross-Cutting Trends

- Repeated themes
- Common methods
- Shared limitations or hype signals

Writing Rules

  • Keep the digest skimmable
  • Prefer short paragraphs and flat bullets
  • Separate "what it says" from "whether it is worth deeper reading"
  • Be explicit when a conclusion is based only on --brief
  • Be explicit when a conclusion is based on --head or a section read
  • Do not pretend to have verified claims you have not checked

Decision Rules

  • If briefs are enough for the user's request, stop there
  • Use --head only for the most promising papers
  • Use --section only after --head suggests a high-value section
  • Do not read full paper markdown unless the user explicitly asks for a full analysis

Minimal Example

deepxiv trending --days 7 --limit 5 --json
deepxiv paper 2603.20639 --brief
deepxiv paper 2603.26221 --brief
deepxiv paper 2603.20639 --head
deepxiv paper 2603.20639 --section Introduction

Then write the digest as markdown and clearly label:

  • all reviewed papers
  • which papers were only briefed
  • which papers received deeper inspection

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.