Abstract

Browser and computer-use agents act on screenshots, DOM snapshots and accessibility trees. Between the moment an agent sees a “Save” button and the moment an external system changes, many things can differ:

  • an overlay or layout shift can send the click to another element;
  • the element may sit in a third-party iframe with a different origin;
  • a redirect can change the destination or method;
  • a service worker or timer can send requests no one clicked;
  • a delayed submission can carry a changed form value;
  • an HTTP 200 or a success banner may not mean the remote transaction committed;
  • a retry after an uncertain result can duplicate a payment.

This disclosure describes a vendor-neutral, non-actuating method for correlating an original authorised task with the browser evidence and the external effect it actually produced:

  • Each task carries a stable identity, principal, allowed effect and approval version.
  • Observations are qualified by DOM version, viewport and focus, and the intended control is re-checked by a fresh hit-test against the element that actually received the event.
  • Frame identity, origin, navigation and redirect chains are reconstructed, so being on a page never confers authority.
  • All effect-producing operations, including those started by workers, timers and APIs, are tied back to the original task.
  • The actual method, destination, payload and idempotency key are compared with what was approved, under current authority and path coverage.
  • External completion is established only from authoritative evidence, and retries are permitted only when the prior effect status is proven.

Twelve expected-versus-actual comparisons and further mandatory predicates are combined in three-valued logic. A contradiction yields INCONSISTENT and is never cancelled by missing evidence; full support yields CONSISTENT; gaps yield UNRESOLVED.

The disclosure provides:

  • a 48-stage, six-phase workflow;
  • 48 technical invariants;
  • six embodiments;
  • 43 formal relations with conditional proofs;
  • 40 checkpoints;
  • a comparison with WebDriver and CDP automation, Playwright actionability checks, DOM and accessibility APIs, same-origin and sandbox policies, CSP and Fetch Metadata, HTTP idempotency keys, saga and outbox patterns, and agent guardrails;
  • a non-actuating evaluator in Python, JavaScript and Go with 74 shared synthetic vectors, all passing (about 0.8 to 3 microseconds median per in-memory evaluation).

It is not a browser engine or an interception product. The laboratory is synthetic, self-authored and not peer reviewed.

A companion industry-relevance section aligns the method with public roadmaps:

  • Google: Gemini in Chrome, auto browse and WebMCP.
  • Microsoft: Copilot Mode in Edge.
  • Other agentic-browser providers: OpenAI, Anthropic (Claude for Chrome), Perplexity (Comet), Opera, Brave and Manus.
  • Chinese computer-use agent developers: Zhipu AI (AutoGLM, AutoClaw), ByteDance (Coze, Volcano Engine ArkClaw), Alibaba (Quark, Agent Bay), Tencent (WorkBuddy) and Baidu AI Cloud.

Disclaimer: The industry and product-roadmap relevance is an AI-generated alignment based only on public announcements, press coverage and published research. It is not an endorsement, partnership, sponsorship, licence, technical validation, security assessment or evaluation by any company, browser vendor or standards body named. It does not state or imply that any company uses, needs, has assessed or infringes this method or any patent application. All company, product and standard names are trademarks of their respective owners and are used only to describe technical relevance. Patent pending: Indian Patent Application No. 202631129178 (source disclosure on Zenodo, record 23793543). Related PCT application: PCT/IB2026/055615, filed 4 June 2026 and publicly disclosed 16 July 2026. Readers must read the original disclosures before drawing any conclusion. The inventor is open to strategic collaboration or acquihire discussions.

Keywords

AI safety, browser agents, computer-use agents, agentic browsers, browser automation, DOM, accessibility tree, hit testing, TOCTOU, iframes, same-origin policy, redirects, service workers, form submission, idempotency, retry safety, external effect verification, prompt injection, three-valued logic, WebDriver BiDi, Chrome DevTools Protocol, WebMCP, Gemini in Chrome, Edge Copilot Mode, Claude for Chrome, Perplexity Comet, Zhipu AutoGLM, ByteDance Coze

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial 4.0 License

Download

Share

COinS