paperslisted
Find, read and compare scientific papers without spending the conversation's context on them: a shortlist of identifiers for a topic, a bounded digest of one paper by a reading subagent, a comparison of several, and a note in the user's vault only when they say so. Sources: the paper-search MCP server or its CLI (arXiv, PubMed, Semantic Scholar, OpenAlex, Crossref and more). Use when the user asks to find papers on a topic, look up a DOI or arXiv id, summarise or critique a scientific article, compare papers, or write a literature note. Not for a web page, a blog post or a news article, which is a plain fetch. Triggers: papers, paper, preprint, arXiv, DOI, PubMed, literature review, systematic review, meta-analysis, related work, citations, who cites this, state of the art, what does the research say, научные статьи, найди статьи, препринт, обзор литературы, мета-анализ, литература по теме, найди источники, есть ли работы про, что говорят исследования, разбери статью, конспект научной статьи, сравни статьи
rokokol/papers-skill · ★ 0 · AI & Automation · score 75
Install: claude install-skill rokokol/papers-skill
# papers
Scientific papers are long and PDFs are worse: the text of one paper is a large fraction of a conversation's context, and a survey is several of them. This skill keeps the reading out of the main conversation. The master finds and decides; a subagent with its own context reads and returns a digest of bounded length; the master synthesises across digests and, only when the user says so, files one as a note
## Setup
- **Sources** come from [paper-search-mcp](https://github.com/openags/paper-search-mcp) in one of two forms. As an MCP server registered under a name (assumed `paper-search` below) its tools are `mcp__paper-search__*`; they are deferred, so before the first call run `ToolSearch` with `select:` and the exact names about to be called, in the subagent as well as in the master: the server has 57 tools and a keyword query returns only the first few, so the one needed is often not among them. As a CLI, `paper-search search|download|read|sources` prints JSON or text. Prefer the MCP where it is registered: `download_with_fallback` and the per-source `read_*_paper` tools exist only there. Check availability once per session: `ToolSearch` finds the tools, or `command -v paper-search` finds the binary; neither means stop and point at [README.md](README.md)
- **Keys** are optional and never pass through the conversation: the server reads an env file named by `PAPER_SEARCH_MCP_ENV_FILE` or `~/.config/paper-search-mcp/.env`. Without them Semantic Scholar shares an ano