Bgpt Tools

Configuration File: bgpt_tools.json Tool Type: Local Tools Count: 1

This page contains all tools defined in the bgpt_tools.json configuration file.

Available Tools

BGPT_search_paper_evidence (Type: BGPTPaperEvidenceTool)

Search scientific papers with BGPT and return structured, full-text-derived study evidence for cr…

BGPT_search_paper_evidence tool specification

Tool Information:

  • Name: BGPT_search_paper_evidence

  • Type: BGPTPaperEvidenceTool

  • Description: Search scientific papers with BGPT and return structured, full-text-derived study evidence for critical appraisal. Unlike PubMed/EuropePMC/OpenAlex (which return titles and abstracts), each result includes methods and experimental techniques, sample size and population, results and conclusions, study context, paper limitations and biases, conflict of interest, data/code availability, software/tools used, quality scores, study blind spots, and a ‘how_to_falsify’ statement. Use it when an agent needs to weigh study quality and evidence before drawing a conclusion (literature review, claim interrogation, hypothesis critique, evidence-dashboard generation). Structured fields are model-generated, so treat them as an appraisal aid rather than curated ground truth. The first 50 results are free; set BGPT_API_KEY for the paid tier afterward.

Parameters:

  • query (string) (required) Natural-language scientific search query, e.g. ‘GLP-1 alcohol craving’ or ‘semaglutide cardiovascular outcomes’.

  • num_results (integer) (optional) Number of paper results to return (1-100, default 10).

  • days_back (integer) (optional) Optional. Restrict results to papers published within the last N days.

  • api_key (string) (optional) Optional paid-tier API key, used after the free result allowance is exhausted. If omitted, the BGPT_API_KEY environment variable is used.

Example Usage:

query = {
    "name": "BGPT_search_paper_evidence",
    "arguments": {
        "query": "example_value"
    }
}
result = tu.run(query)