agentleFS
fritzprixfritzprix/libr-agent/.agents/skills/science-search/SKILL.md

science-search

Skill12 starsChanged 6 days agoMITView on GitHub

Search ~200 million scientific papers (OpenAlex titles and abstracts) by meaning or exact words, look up papers by DOI / OpenAlex ID / PMCID, find similar papers, and walk citation graphs through the free ScienceMadeReadable search API (no key). Use when a task requires scientific literature, finding evidence for claims, conducting literature reviews, citing peer-reviewed studies, or resolving DOIs. Triggers on: 'science search', '논문 검색', '과학 논문', '학술 자료', 'research paper', 'scientific study', 'find paper', 'DOI 검색', '의학 논문', '선행 연구'.

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 1 is distinctive, and 4 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 4

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Search ~200 million peer-reviewed papers and academic preprints across all fields of science.

    line 8
  2. Very specific:

    If you only have a web fetching tool (read_url_content / HTTP GET), request the plain-text Markdown endpoint directly:

    line 20
  3. Very specific:

    Run the bundled script using standard Python (python3, python, or uv run):

    line 31
  4. Very specific:

    Step 1: Search with abstracts=false and k=10~20 to inspect titles, publication years, citation counts, and venues.

    line 119
  5. Very specific:

    Always cite using the official DOI (https://doi.org/<doi>) or OpenAlex URL.

    line 128

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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

Science Search (ScienceMadeReadable)

Search ~200 million peer-reviewed papers and academic preprints across all fields of science.

  • Base URL: https://api.sciencemadereadable.com/api/data
  • Authentication: Free, read-only JSON over HTTPS, no API key required.
  • Reference Doc: GET https://api.sciencemadereadable.com/api/data/llms.txt

1. Quick Invocation (Cross-Platform)

Option A: Direct Web Fetch / Agent Fallback (No CLI Required)

If you only have a web fetching tool (read_url_content / HTTP GET), request the plain-text Markdown endpoint directly:

https://ask.sciencemadereadable.com/<question>
  • Spaces can be %20 or + (e.g. https://ask.sciencemadereadable.com/why%20do%20cats%20purr?k=10&abstracts=false)
  • Accepts standard query parameters: k (1–100), year_from, year_to, type, open_access=true, abstracts=false.

Option B: Bundled Python Helper Script (Zero Dependencies)

Run the bundled script using standard Python (python3, python, or uv run):

# Semantic search with token-saving (no abstracts)
python scripts/search_science.py "effect of intermittent fasting on weight loss" --limit 10 --no-abstracts

# Keyword / Boolean search
python scripts/search_science.py '"gut microbiome" AND autism NOT mice' --limit 5

# Look up paper by DOI or OpenAlex ID
python scripts/search_science.py --work "10.1126/science.aan8048"

# Find similar papers
python scripts/search_science.py --similar "W2782456798" --limit 5

# Citation walk (what similar papers cite that this paper does not)
python scripts/search_science.py --walk "10.1126/science.aan8048"

Option C: cURL / Shell

Bash (Linux / macOS):

B="https://api.sciencemadereadable.com/api/data"

# Semantic search
curl -sG "$B/search" --data-urlencode "q=effect of intermittent fasting on weight loss in adults" -d k=10 -d abstracts=false

# Keyword & boolean search
curl -sG "$B/search" --data-urlencode 'q="gut microbiome" AND (autism OR ASD) NOT mice' -d k=10 -d abstracts=false

# Paper lookup by DOI
curl -s "$B/works/10.1126/science.aan8048"

# Similar papers
curl -s "$B/similar/W2782456798?k=10&abstracts=false"

# Citation walk
curl -s "$B/walk?id=10.1126/science.aan8048"

PowerShell (Windows):

$B = "https://api.sciencemadereadable.com/api/data"

# Semantic search
Invoke-RestMethod -Uri "$B/search?q=effect+of+intermittent+fasting+on+weight+loss+in+adults&k=10&abstracts=false"

# Paper lookup
Invoke-RestMethod -Uri "$B/works/10.1126/science.aan8048"

2. API Endpoints Reference

EndpointMethodPurposeKey Parameters
/searchGETSearch papers by meaning or exact keywordsq, mode (auto|semantic|keyword), k (1–100), year_from, year_to, type, open_access, abstracts (true|false)
/search/{question}GETPlain-text Markdown output of search resultsSame query options as /search
/works/{ref}GETFull paper record by DOI, W-number, or PMCIDref = DOI (10.xxx), OpenAlex ID (Wxxx), or PMCID (PMCxxx)
/similar/{ref}GETFind papers closest in meaning to a specific paperref, k, year_from, year_to, abstracts
/walk?id={ref}GETCitation walk: find literature that similar papers citeid (DOI or W-number), k

Key Result Fields

  • id: OpenAlex W-number (e.g. W2782456798)
  • doi: Digital Object Identifier
  • title: Paper title
  • year / date: Publication date
  • authors: Authors list (joined by " | ")
  • venue / publisher: Journal or conference name
  • citations: Citation count at snapshot date
  • abstract: Paper abstract (if abstracts=true)
  • score: Cosine similarity (semantic) or BM25 score (keyword)
  • junk: Likelihood (0.0–1.0) that the item is spam or non-scholarly
  • openalex: OpenAlex URL for citation

3. Best Practices for AI Agents

  1. Academic Phrasing Optimization:
    • Transform informal questions into academic/scientific phrasing before searching.
    • Example: Instead of "how to cure insomnia", search for "non-pharmacological interventions for chronic insomnia in adults". This drastically increases semantic matching precision.
  2. Token Efficiency (Screening Pattern):
    • Step 1: Search with abstracts=false and k=10~20 to inspect titles, publication years, citation counts, and venues.
    • Step 2: Select the top 1~3 most relevant papers and call /works/{id} or re-query with abstracts=true only for those specific papers.
  3. Snowballing Literature Reviews:
    • Identify 1 seminal paper $\rightarrow$ call /similar/{id} to discover peer research $\rightarrow$ call /walk?id={doi} to capture foundational works cited across the field.
  4. Rate Limits & Etiquette:
    • Limit: 30 requests per minute per IP address.
    • Batch parameters into a single request (e.g. k=30 once rather than multiple k=5 calls).
    • Respect HTTP 429: read the Retry-After header or retry_after JSON response field and sleep before retrying.
  5. Citing Papers:
    • Always cite using the official DOI (https://doi.org/<doi>) or OpenAlex URL.
    • To guide human users to an interactive reading view: https://sciencemadereadable.com/?work=<doi>.

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.