pubmed-edirect

What This Skill Does

The pubmed-edirect skill enables OpenClaw to perform deep queries across the vast NCBI database ecosystem. By leveraging the official Entrez Direct (EDirect) CLI utilities, this skill allows the agent to search, retrieve, filter, and extract complex biological and medical literature from PubMed, PMC, and other NCBI-hosted databases. It acts as a bridge between the OpenClaw execution environment and local, high-performance bioinformatics tools, providing the AI with the ability to programmatically access millions of scientific articles and research records directly from the command line.

Installation

Installation is a multi-step, security-conscious process. First, you must install the OpenClaw skill via clawhub install openclaw/skills/skills/killgfat/pubmed-edirect. After the skill infrastructure is present, you must perform a manual installation of the underlying EDirect binaries. Download the official installer script from the NCBI FTP server, review the code to ensure it matches the verified source, and execute it locally. Finally, update your shell PATH to include the EDirect executables. We recommend validating the installation by running esearch -db pubmed -query "cancer" -retmax 1 in your terminal to ensure the environment is correctly configured.

Use Cases

Researchers can use this skill to automate literature reviews, aggregate abstracts for meta-analysis, or track emerging trends in specific biological fields. Developers may utilize it to extract gene, protein, or nucleotide metadata from NCBI records for training or data enrichment. It is perfect for users who need to perform complex, repetitive queries that are otherwise time-consuming or impossible via standard web interfaces.

Example Prompts

1. "Search PubMed for the latest 5 reviews on CRISPR-Cas9 gene editing published in the last month and summarize the findings."
2. "Find all gene entries for BRCA1 and extract the official symbol and primary summary using the xtract utility."
3. "Retrieve the full text metadata for PMC article ID 7482937 and save it to a local markdown file."

Tips & Limitations

Note that this is not a containerized skill; it runs directly on your host machine. Always verify the permissions of the scripts you execute. Since PubMed data can be voluminous, use the -retmax flag to control result sizes and avoid network timeouts or system strain. When piping multiple commands (e.g., esearch | efetch | xtract), monitor your memory usage. If you are dealing with very large datasets, consider filtering your initial search queries extensively before fetching full metadata.