More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    pubmed systematic review

    3

    Run reproducible, PRISMA-compliant PubMed searches with batch fetching, metadata exports, and open-access verification.

    Free

    6 uses5.0 (2 reviews)
    pubmed systematic review

    pubmed systematic review

    Example session with this skill installed

    Search PubMed for "long covid AND exercise intolerance", create a snapshot, and generate a CSV export of the first 50 results.

    • Read your context and instructions
    • Compiled the pubmed systematic review

    I completed a provenance-first PubMed search for your systematic-review style query. Nothing was inferred — only fields returned by NCBI are present, and missing values stay explicit.

    Search summary

    MetricValue
    Original querylong covid AND exercise intolerance
    Query translation("long covid"[All Fields] AND ("exercise intolerance"[All Fields] OR ("exercise"[MeSH Terms] OR "exercise"[All Fields])))
    Total matches142
    Retrieved for export50 (first page; full set available via history)
    Snapshot IDPUBMED-20261017-A3F9C2
    Export formatCSV
    Anti-inference checkfalse_inference_count = 0

    CSV preview (first 5 of 50)

    pmid,title,authors,journal,pub_date,doi,publication_types,url
    37841209,Exercise Intolerance in Post-Acute Sequelae of COVID-19,"Smith J; Doe A",J Am Coll Cardiol,2023 Oct,10.1016/j.jacc.2023.08.012,Journal Article,https://pubmed.ncbi.nlm.nih.gov/37841209/
    37523145,Pathophysiology of exercise intolerance in long COVID,"Miller R; et al.",J Transl Med,2023 Aug,10.1186/s12967-023-04351-y,Review,https://pubmed.ncbi.nlm.nih.gov/37523145/
    37210984,Development of a Definition of Postacute Sequelae of SARS-CoV-2 Infection,"Tanaka S; et al.",JAMA,2023 Jun,10.1001/jama.2023.8823,Journal Article,https://pubmed.ncbi.nlm.nih.gov/37210984/
    36982140,Long covid: exercise and rehabilitation,"Greenhalgh T",BMJ,2023 Mar,10.1136/bmj-2022-072117,Clinical Conference,https://pubmed.ncbi.nlm.nih.gov/36982140/
    36891022,The cardiovascular system and long COVID,"Davis H; et al.",Nat Rev Cardiol,2023 Apr,10.1038/s41569-023-00842-w,Review,https://pubmed.ncbi.nlm.nih.gov/36891022/
    

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Building a systematic review requires precision that generic LLMs lack. This skill solves the problem of "hallucinated" citations and non-reproducible searches by connecting your agent directly to the NCBI and Europe PMC databases. It ensures every piece of metadata, from DOIs to sample sizes, is pulled via official APIs with strict provenance tracking.

    What it does

    • Reproducible searching executes Boolean and MeSH queries, capturing the exact NCBI query translation and total match counts.
    • Batch retrieval handles large result sets up to 9,999 records using NCBI history servers to avoid data loss.
    • Search snapshots generates durable snapshot IDs for PRISMA-compliant documentation and later export.
    • Format export converts search results into RIS, BibTeX, CSV, or JSON formats for reference managers like Zotero or EndNote.
    • Open access verification checks legal PDF availability via Unpaywall and Europe PMC, including secure downloads of open-access papers.
    • Systematic deduplication merges records across PubMed and Europe PMC using PMID, DOI, and PMCID mapping.

    How it works

    1. Search using pubmed_search with Boolean strings to retrieve PMIDs and capture the webenv history key.
    2. Fetch full records in batches using pubmed_batch_fetch to ensure no studies are skipped due to page limits.
    3. Snapshot the results with pubmed_create_search_snapshot to create a permanent record of the search state.
    4. Analyze using pubmed_evidence_table to generate a structured overview of study designs and sample sizes without inventing data.

    Frameworks & tools

    This tool requires the pubmed-access-mcp server. It utilizes the NCBI Entrez API, Europe PMC API, and Unpaywall API. Access requires NCBI_EMAIL, while NCBI_API_KEY is recommended for higher rate limits.

    Why this beats prompting it yourself

    Standard LLMs frequently hallucinate PMIDs or misquote study findings when relying on training data. This skill enforces a "provenance-first" rule, where the agent is physically unable to invent metadata and must explicitly label missing fields, ensuring your review meets scientific standards.

    Use cases

    • Generating PRISMA-compliant search logs for meta-analysis manuscripts.
    • Exporting large batches of citations to RIS format for screening in Covidence.
    • Verifying the retraction status of a list of PMIDs before publication.
    • Automating the collection of legal, open-access PDFs for a research team.

    Known limitations

    NCBI limits ESearch to approximately 9,999 UIDs; larger sets require date partitioning. It cannot bypass paywalls or grade evidence quality (GRADE/CEBM) from abstracts alone.

    NCBI ESearch limit of 9,999 UIDs per query. Requires NCBI_EMAIL environment variable. Cannot grade evidence quality from abstracts alone. No paywall bypassing.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    5.0
    2 reviews
    5
    2
    4
    0
    3
    0
    2
    0
    1
    0

    This skill has been used 6 times.

    Trust & safety

    Security scan

    Security scan passed 1 month ago

    • Free with an Agensi account

    Requires

    Pubmed Access MCP
    uv (Python tool runner)

    Permissions

    Terminal / Shell
    Network Access
    Listed1 month ago

    What's inside

    Frequently Asked Questions

    Unlike generic LLMs that may hallucinate PMIDs or study details, this skill connects directly to official NCBI and Europe PMC APIs, ensuring every citation, DOI, and sample size is retrieved with 100% accuracy and strict provenance tracking.

    This skill is designed for agents compatible with the Model Context Protocol (MCP); it requires the pubmed-access-mcp server and works with popular LLM frameworks that support tool-calling and external API integration.

    You will need to provide an NCBI_EMAIL for basic access, and while not strictly required, an NCBI_API_KEY is highly recommended to enable higher rate limits for large batch fetches and faster data retrieval.

    Yes, the skill includes a `pubmed_create_search_snapshot` tool that captures exact NCBI query translations and match counts, providing the reproducible documentation required for PRISMA-compliant manuscripts.

    The skill can retrieve full metadata for up to 9,999 records per search using NCBI history servers and offers batch export options in RIS, BibTeX, CSV, and JSON formats for seamless use with Zotero or EndNote.

    While it cannot bypass publisher paywalls, the skill integrates with Unpaywall and Europe PMC to automatically verify legal open-access status and securely download available PDFs for your research team.

    New in Research & Analysis