Code & Chain · Signal Desk

OpenAI Adds Annotation Feature to WebMCP So Browser Agents Can Read Page Elements

Original sourceJames Okafor

Summary

OpenAI has extended WebMCP with an annotation feature that lets browser agents directly see specific marked elements on a page without relying on parsing raw HTML or screenshots. Announced on October 12, 2026, the feature provides precise spatial and semantic context, aiming to improve automation r…

Key points

  • The bottleneck for browser agents is often understanding page structure, and precise annotations can improve automation success rates and cut debugging costs.
  • Providing precise annotations for page elements is a key part of moving browser agents from experimentation to practical automation.
  • Teams building browser agents can switch to page annotations for more stable element location, but given the lack of independent performance data, they should test it themselves before adopting.

Editorial note

This page is Code & Chain's editorial summary of public sources. It may be prepared with AI assistance and published through an automated workflow. Refer to the original sources; this content is not investment, legal, or tax advice.