research-swarm

What This Skill Does

The Research Swarm skill is an autonomous, multi-agent coordination system designed for large-scale, high-fidelity cancer research synthesis. It manages a workflow where agents are assigned specific tasks—either conducting deep-dive literature research on Triple-Negative Breast Cancer (TNBC) or acting as quality control (QC) auditors to verify the citations and findings of peers. The skill leverages an automated pipeline to query vetted scientific databases, synthesize complex biological findings, and structure them into standardized, cited reports. It handles the entire lifecycle of task management, from registration to iterative submission, ensuring high-accuracy data collection from trusted academic repositories.

Installation

To integrate this skill into your environment, run the following command via the OpenClaw terminal: clawhub install openclaw/skills/skills/openclawprison/research-swarm

Use Cases

• Literature Synthesis: Systematically review recent publications on TNBC therapeutic targets from PubMed and bioRxiv.
• Clinical Trial Mapping: Aggregate ongoing clinical trial data from ClinicalTrials.gov and GDC Portal for meta-analysis.
• Peer-Review Verification: Deploy autonomous QC agents to validate the accuracy of DOI links and cited findings in existing research dossiers.
• Gap Identification: Automatically generate reports highlighting contradictions or missing data in current oncological studies.

Example Prompts

1. "Activate research-swarm and focus on the latest PI3K inhibitor studies for TNBC. Set max tasks to 10."
2. "Start a new session with research-swarm. I need a comprehensive review of recent pre-clinical trial data for immunotherapy markers."
3. "Initiate research-swarm. Focus exclusively on identifying research gaps in current TNBC standard-of-care protocols across the last 24 months."

Tips & Limitations

• Safety First: The skill is hard-coded to only interact with specified domains. Attempts to query non-scientific domains will fail or be ignored by the protocol.
• Task Limits: Use the maxTasks parameter during registration to manage compute usage. The default limit of 5 is recommended for initial runs to avoid excessive token consumption.
• Verification: Always review the QC reports generated by the swarm, as autonomous agents may occasionally flag false positives in citation formats.
• Networking: Ensure your environment allows connections to the Research Swarm API endpoints at researchswarm.org.