agentleFS
aws-samplesaws-samples/amazon-bedrock-agents-healthcare-lifesciences/skills/research-deep-literature-review/SKILL.md

research-deep-literature-review

Skill277 starsChanged 6 months agoMIT-0View on GitHub

Use when a developer needs to conduct systematic literature searches, find research papers, extract content from publications, or build evidence summaries from biomedical databases and web sources via the Biomni Research Tools MCP server.

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 8 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈8

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    All tools accept natural language prompts.

    line 43
  2. Distinctive:

    Cross-reference: Check multiple databases for consistency

    line 59
  3. Distinctive:

    Extract details: Use PDF extraction tool for full-text papers

    line 60
  4. Distinctive:

    Extract details: Use PDF extraction for advisory briefing documents

    line 68
  5. Distinctive:

    Validate: Check population frequencies (gnomAD) for genomic findings

    line 75
  6. Distinctive:

    Be specific with organisms: Always specify "human" to avoid cross-species results

    line 86
  7. Distinctive:

    Cross-validate: If a finding appears across multiple database tools, it's more reliable

    line 89
  8. Distinctive:

    Extract strategically: Only use PDF extraction when you need methods details or specific data points

    line 90

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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

Deep Literature Review

When to use this skill

  • Developer asks "find papers about X" or "what's the latest research on Y?"
  • Developer needs to build an evidence summary from multiple publications
  • Developer wants to extract specific data from a paper or PDF
  • Developer needs to search across biomedical literature and databases
  • Developer asks about systematic literature review workflows

MCP Server: biomni-research

This server provides tools for biomedical database queries including literature-related functionality. Tools are discovered automatically via the MCP protocol.

Quick setup (after deployment):

cd mcp-servers/agentcore-gateway/biomni-research-tools
source get-token.sh
claude mcp add --transport http biomni-research "$BIOMNI_GATEWAY_URL" --header "Authorization: Bearer $BIOMNI_MCP_TOKEN"

Search Strategy

Choosing the Right Approach

Research TypeQuery approach (via biomni-research)
Clinical evidence (trials, outcomes)Query ClinicalTrials database
Variant significanceQuery ClinVar + gnomAD
Drug safety / regulatoryQuery OpenFDA
Protein biologyQuery UniProt + InterPro
Cancer genomicsQuery cBioPortal + GEO
Pathway mechanismsQuery Reactome + Open Targets

Effective Query Construction

All tools accept natural language prompts. Be specific:

"Find pathogenic BRCA1 variants associated with breast cancer"
"EGFR T790M resistance mutation frequency in Asian populations"
"CDK4/6 inhibitors clinical trials phase 3 breast cancer"
"TP53 expression in hepatocellular carcinoma datasets"

Literature Review Workflows

Workflow 1: Rapid Evidence Summary

Goal: Quickly assess the state of evidence on a topic.

  1. Broad database search: Query relevant databases for the topic
  2. Cross-reference: Check multiple databases for consistency
  3. Extract details: Use PDF extraction tool for full-text papers
  4. Synthesize: Combine database evidence with extracted findings

Workflow 2: Competitive Intelligence (Drug Development)

  1. Find clinical trials: Query ClinicalTrials for "drug/target phase 2 phase 3 results"
  2. Check target biology: Query Open Targets + UniProt for target validation
  3. Regulatory context: Query OpenFDA for drug safety signals
  4. Extract details: Use PDF extraction for advisory briefing documents

Workflow 3: Systematic Literature Search

  1. Define PICO: Population, Intervention, Comparison, Outcome
  2. Multi-database search: Query ClinVar + ClinicalTrials + Open Targets
  3. Cross-reference: Compare findings across databases for consistency
  4. Validate: Check population frequencies (gnomAD) for genomic findings

Workflow 4: Technology Landscape

  1. Foundational biology: Query UniProt + Reactome for mechanism
  2. Clinical translation: Query ClinicalTrials for applications
  3. Cancer data: Query cBioPortal for mutation/expression profiles
  4. Regulatory status: Query OpenFDA for approved therapies

Tips

  • Be specific with organisms: Always specify "human" to avoid cross-species results
  • Use standard identifiers: Gene symbols (BRCA1), UniProt IDs (P38398), RS numbers (rs80357906)
  • Start broad, narrow down: First query identifies the entity → follow-up gets specific data
  • Cross-validate: If a finding appears across multiple database tools, it's more reliable
  • Extract strategically: Only use PDF extraction when you need methods details or specific data points

Combining with Other Skills

Literature review is most powerful when combined with other database queries:

  1. Find variant: Query ClinVar for "EGFR T790M pathogenic"
  2. Check frequency: Query gnomAD for "EGFR T790M population frequency"
  3. Check pathways: Query Reactome for "EGFR signaling cascade"
  4. Find trials: Query ClinicalTrials for "EGFR T790M osimertinib"

This multi-database feedback loop produces comprehensive evidence packages.

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.