Paper Search
Search 250M+ academic papers via OpenAlex for free — keyword search, DOI lookup, sort by citations or date, no API key needed.
Read-only scholarly search via the free OpenAlex API, no API key, clear data flow (curl+jq returning metadata); no sensitive-data collection or dangerous side effects observed. However, the actual search.sh/paper.sh sources are absent from the evidence, so least privilege, error handling, and outbound request scope cannot be statically verified; publisher identity is unverified. Deducted accordingly.
SKILL.md call paths and parameters (limit/sort/page) are consistent with README examples and dependencies (curl, jq) are declared; happy path is plausible. But script bodies are unseen, abnormal-input handling and failure feedback are unverifiable, and locating scripts via `find ~/.claude ... tail -1` is fragile across versions, so the static cap of 10 applies.
Scenario is clear (keyword search, DOI/ID lookup, citation/date sort) with precise triggers; but capability boundaries and non-fit ranges (OpenAlex coverage only, possible missing abstracts) are undeclared, and the core function depends entirely on the overseas OpenAlex service, with mainland-China reachability uncertain though typically available. Partial deduction.
Docs are readable with install steps, dependencies, usage examples and API notes; MIT license is complete. But there is no versioning, no changelog, limited known-limitation disclosure (rate limits, abstract coverage), and maintenance responsibility is only implicit in a personal repo. Deducted.
The claimed value (free, key-less search with citation counts and open-access links) offers clear marginal benefit for researchers/writers, but static review executed nothing; output format correctness and direct usability are unverified, and missing script bodies leave core promises as claims only. Deducted.
Auditable primary material (SKILL.md, README, LICENSE) exists and cross-references the external OpenAlex service, but there are no tests, no CI execution evidence, and no sample outputs; conclusions rest largely on self-description, so under static review the score stays below 5.
- The search.sh and paper.sh script sources were not included in the reviewed evidence; actual behavior (request scope, error handling) is unverified — manually review the scripts before installing.
- The skill depends entirely on the overseas OpenAlex service; mainland-China network reachability may vary by environment — test connectivity yourself.
- No versioning or changelog; behavior may change on update, and the `find ... tail -1` script-locating approach may pick the wrong version when multiple installs exist.
- Abstracts showing 'N/A' reflect OpenAlex coverage limits; use paper.sh per the skill's own tips to fetch details.
What does this skill do, and when should you use it?
paper-search is a Claude Code plugin and skill that wraps the OpenAlex API in simple shell scripts. It searches papers by keyword, looks up individual papers by DOI or OpenAlex ID, and sorts results by relevance, citation count, or publication date. Results include title, authors, citation count, DOI, abstract, and open access links. It is completely free and requires no registration or API key.
The skill body instructs the model to locate the installed scripts directory, then call two bash scripts. search.sh takes a query, a per-page limit (default 10, max 200), a sort mode (relevance/cites/date), and a page number, and requests paper lists from OpenAlex. paper.sh takes a DOI URL or OpenAlex ID and returns full details for one paper, including authors, abstract, concepts, an open access PDF link, and related works. It depends on curl and jq and requires network access.
- A writer who needs scientific backing for claims: search broadly, pick the most relevant or most cited papers, then use paper.sh for full abstracts and cite as (Author, Year, Journal)
- A literature reviewer: sort by cites to quickly surface landmark papers on a topic
- Anyone with a DOI in hand who wants a quick overview: paper.sh returns authors, abstract, and open access PDF in one step
- A researcher tracing citation networks: feed related_works IDs from paper.sh back into paper.sh to explore the citation graph
- Someone tracking recent work: sort by date to see the newest publications on a topic
What are this skill's strengths and limitations?
- Completely free, built on open OpenAlex data, no API key or sign-up
- Indexes 250M+ scholarly works
- Supports both keyword search and exact DOI/ID lookup, with pagination and three sort modes
- Returns open access PDF links for direct full-text access
- MIT licensed and open source
- Implemented with plain bash + curl + jq — minimal dependencies, easy to audit
- Requires curl and jq; it will not run without them
- Single data source (OpenAlex); does not cover other scholarly databases
- Search results sometimes show 'Abstract: N/A', requiring an extra paper.sh call
- No test suite, error handling, or rate-limit guidance is documented in the source
How do you install this skill?
Run in a terminal:
claude plugin marketplace add ykdojo/paper-search
claude plugin install paper-search@paper-search
After installing, the paper-search skill is available automatically. Make sure curl and jq are installed. The source does not document manual (non-plugin) folder placement steps beyond noting the skill can also be found under ~/.claude when installed manually.
How do you use this skill?
Once the plugin is installed, just ask Claude Code to search for papers and it will use the skill. You can also run the scripts directly:
scripts/search.sh "mindfulness meditation stress reduction" 10 cites # sort by citations
scripts/search.sh "query" 10 relevance 2 # page 2 of results
scripts/paper.sh "https://doi.org/10.1073/pnas.0407162101" # look up by DOI
scripts/paper.sh W2097529540 # look up by OpenAlex ID