Paper Analysis
Routes a paper to a benchmark, method or survey framework for a deep read
Test report
- Verdict
- Tested · Works
- Score
- Tested
- Aug 5, 2026
- Environment
- Claude Code 2.x (agent harness)
⚠ This skill is no longer available upstream. Our re-check on Aug 10, 2026 couldn't find it any more (repo unreachable/deleted). The test below is what we measured on Aug 5, 2026 and we're leaving it up as a record — but there is nothing left to install, so we've removed the command.
Analyzed the SWE-bench paper twice, once cold and once through the skill's benchmark framework, and the framework earned its length: it forced a trip to the Hugging Face dataset card that produced the exact twelve-field schema with FAIL_TO_PASS and PASS_TO_PASS semantics, and surfaced that the published test split is 2,290 rows against the 2,294 the paper claims — a discrepancy my unaided summary repeated without noticing. It also demanded a LaTeX formalization of percent-resolved that makes the all-or-nothing scoring explicit, an evaluation-cost estimate that explains why the community migrated to the Lite and Verified subsets, and author-lineage context the baseline skipped entirely. Its stated degradation rule fired for real when the swebench.com leaderboard fetch came back truncated, so the leaderboard trend went out labelled unverified instead of invented. The costs are real: output is roughly four times longer, hard-coded to Chinese with no language switch, and the mandated citation-network tracing is more than one context can finish. The README's install line is 'npx skills add' into ~/.cursor/skills, so Claude Code users have to derive the copy command themselves.
Scored on four weighted criteria — install, triggering, output vs. baseline, docs. How scoring works
- Installs cleanly 5/5
- Triggers reliably 5/5
- Output vs. baseline 8/10
- Docs & honesty 4/5
What Paper Analysis does
Classifies an AI research paper as a benchmark, methodology or survey/position piece and loads the matching nine-section analysis framework, with an extra layer for Nature/Science/Cell journal papers. Triggers when you ask to read, analyze or review a paper, or use the Chinese terms 论文分析 and 论文阅读. Output is written in Chinese and relies on the agent's own web fetch, with no paid scraping service required.
How to install Paper Analysis
Nothing to install: the source repository no longer has this skill. If the author brings it back, our daily re-check will pick it up and the command will reappear here.
Commands — how to trigger Paper Analysis
-
/paper-analysisRoutes a paper to a benchmark, method or survey framework for a deep read
It also activates on plain-language prompts like these:
-
请帮我从头到尾精读一遍这篇论文，详细分析它的研究方法、创新点、实验设计和主要局限性 -
深入分析一下这篇 arXiv 论文 2310.06770，看看它的实验设计是否严谨、数据是否可信 -
这是一篇 benchmark 类型的论文，麻烦按照标准框架帮我做一次系统而深入的论文分析
Frequently asked questions
- Is the Paper Analysis skill free?
- Yes. The skill itself is free from Ayanami0730/paper-read-skill. SkillProof publishes the install command and an independent test verdict at no cost.
- Does Paper Analysis work with Claude Code?
- We tested it with Claude Code 2.x (agent harness) on Aug 5, 2026. Verdict: Tested · Works. Analyzed the SWE-bench paper twice, once cold and once through the skill's benchmark framework, and the framework earned its length: it forced a trip to the Hugging Face dataset card that produced the exact twelve-field schema with FAIL_TO_PASS and PASS_TO_PASS semantics, and surfaced that the published test split is 2,290 rows against the 2,294 the paper claims — a discrepancy my unaided summary repeated without noticing. It also demanded a LaTeX formalization of percent-resolved that makes the all-or-nothing scoring explicit, an evaluation-cost estimate that explains why the community migrated to the Lite and Verified subsets, and author-lineage context the baseline skipped entirely. Its stated degradation rule fired for real when the swebench.com leaderboard fetch came back truncated, so the leaderboard trend went out labelled unverified instead of invented. The costs are real: output is roughly four times longer, hard-coded to Chinese with no language switch, and the mandated citation-network tracing is more than one context can finish. The README's install line is 'npx skills add' into ~/.cursor/skills, so Claude Code users have to derive the copy command themselves.
- What is the Paper Analysis SkillProof Score?
- 8.8/10 — installs cleanly 5/5, triggers reliably 5/5, output vs. baseline 8/10, docs & honesty 4/5.
- How do I install Paper Analysis?
- Copy the install command from this page, run it in your terminal, and restart Claude Code. Skills live in ~/.claude/skills/ (global) or .claude/skills/ inside a project.
- Can I use Paper Analysis with Cursor, Copilot, Gemini CLI, Codex or other AI tools?
- The SKILL.md format is native to Claude (Claude Code, Desktop, claude.ai). The instructions inside adapt to other assistants: Cursor rules, GitHub Copilot instructions, Windsurf rules, Custom GPTs, AGENTS.md for OpenAI Codex, and GEMINI.md for Google Gemini CLI — our conversion guides cover each, and the free converter on the tools page does the wrapping for you.