web-verification

---

name: web-verification description: Grounds clinical diagnosis via live medical literature search using Google Search Tool license: MIT compatibility: opencode metadata: audience: developers workflow: clinical-pipeline

What I do

I ground the AI-generated diagnosis in real medical literature by searching for relevant sources. I verify that the diagnosis aligns with current clinical guidelines and research.

When to use me

Use this when:

• Fairness validation is complete and you need web verification
• You need to cite authoritative sources for the diagnosis
• You're ensuring the diagnosis aligns with current medical knowledge

Key Concepts

• Google Search Tool: Live web search for medical literature
• Source Verification: Confirm diagnosis with clinical guidelines
• Web Evidence: Citations from peer-reviewed sources
• web_verified: State flag after verification complete

Source Files

• services/router.ts: Web verification routing
• types.ts: AnalysisResult with webVerification field

Code Patterns

• Search for relevant medical literature based on detected conditions
• Verify diagnosis aligns with clinical guidelines
• Return verified flag, sources, and summary

Operational Constraints

• Must provide at least one authoritative source
• Non-critical failure - return "skipped" rather than crash
• Flag AI reasoning vs. grounded web sources in output