Dev.to
6/20/2026

No agent calls WebMCP yet. Here's the honest case for shipping it today — and how to know the day that changes.
Short summary
WebMCP isn't wired into any mainstream AI agents yet (ChatGPT, Claude, Gemini all still scrape DOM), but shipping low-cost, feature-detected tools today is strategically sound: the cost is minimal (one afternoon, no migration debt), and instrumentation from day one lets you see exactly when agents arrive. Feature-detect navigator.modelContext, reuse existing handlers, log tool calls—so you capture value the moment adoption tips.
- •WebMCP adoption cost is minimal: feature-detected, additive, zero runtime risk
- •Real value emerges when agents start calling tools—instrumenting early lets you see it
- •Ship with proper instrumentation (beacon logging) to avoid the 'set and forget' failure mode
Generated with AI, which can make mistakes.
Is this a good recommendation for you?



