Skip to main content
Glama

validate_sources

Verify the credibility of sources and claims using predefined validation frameworks such as empirical, responsible, harmonic, or pluralistic methods to ensure ethical and reliable information.

Instructions

Validate sources and evidence using configured framework

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
frameworkNoValidation framework to use
textYesText containing claims and sources to validate

Implementation Reference

  • Main execution logic for the validate_sources tool: parses framework, extracts potential sources via regex matching phrases like 'according to', generates framework-specific validation prompts using external MCP servers (Exa, Brave, ArXiv, Google Scholar), and returns text summary and JSON with sources and prompts.
    if (name === "validate_sources") { const framework: keyof typeof VALIDATION_FRAMEWORKS = (typeof args.framework === 'string' && args.framework in VALIDATION_FRAMEWORKS) ? args.framework as keyof typeof VALIDATION_FRAMEWORKS : VALIDATION_FRAMEWORK; // Extract sources const sourcePattern = /according to|cited by|reported by|study by|research by|experts|scientists/gi; const sources = []; let match: RegExpExecArray | null; while ((match = sourcePattern.exec(args.text)) !== null) { const context = args.text.substring( Math.max(0, match.index - 30), Math.min(args.text.length, match.index + 70) ); sources.push({ type: "citation", context: context.trim() }); } // Generate validation prompts for each source const validationPrompts = sources.flatMap((source: { type: string; context: string }) => { const basePrompts = [ `- Use Exa MCP server to verify credibility of: "${source.context}"`, `- Use Brave Search to find independent verification: "${source.context}"`, `- Search ArXiv for related technical papers: "${source.context}"`, `- Use Google Scholar MCP server to check academic citations: "${source.context}"`, `- Cross-reference findings between different platforms to establish credibility` ]; if (framework === "empirical" || framework === "pluralistic") { basePrompts.push( `- Compare methodologies and results across different studies`, `- Verify replication status and reproducibility`, `- Cross-validate findings between different research groups` ); } if (framework === "responsible" || framework === "pluralistic") { basePrompts.push( `- Use Exa MCP server to search for community perspectives: "${source.context}"`, `- Compare academic findings with real-world impacts`, `- Cross-reference with local knowledge and experiences` ); } if (framework === "harmonic" || framework === "pluralistic") { basePrompts.push( `- Compare perspectives across different cultural contexts`, `- Synthesize findings from multiple knowledge systems`, `- Identify areas of consensus and divergence` ); } return basePrompts; }); return { content: [ { type: "text", text: `Source validation using ${framework} framework:\n\n` + `Found ${sources.length} sources to validate.\n\n` + `Validation steps:\n${validationPrompts.join("\n")}` }, { type: "text", text: JSON.stringify({ framework, sources, validationPrompts }, null, 2) } ], }; }
  • Input schema for validate_sources tool defining required 'text' parameter and optional 'framework' enum.
    { name: "validate_sources", description: "Validate sources and evidence using configured framework", inputSchema: { type: "object", properties: { text: { type: "string", description: "Text containing claims and sources to validate", }, framework: { type: "string", description: "Validation framework to use", enum: ["empirical", "responsible", "harmonic", "pluralistic"], } }, required: ["text"], }, },
  • src/index.ts:20-22 (registration)
    Server capabilities registration declaring support for the validate_sources tool.
    analyze_claim: true, validate_sources: true, check_manipulation: true

Other Tools

Related Tools

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/bmorphism/anti-bullshit-mcp-server'

If you have feedback or need assistance with the MCP directory API, please join our Discord server