Articles tagged "ai-agents"

4 articles found

Four circles in a row: three filled with orange watercolor that fades across them, the fourth an empty hairline ring — request, grant, revoke, and the audit primitive that was never filled in.

Mozilla wants AI agents to inherit the extension permission model — I counted what that model looks like at scale

Mozilla's W3C proposal puts AI agent permissions on the browser's extension model. I counted that model at scale: 300,503 extensions, 12,277 already pulled.

Maxim Kosterin

August 10, 2026

7 min read

A zen enso ink circle with a small gap, from which a single orange watercolor thread escapes into empty space — a sealed boundary that quietly leaks.

browser-extensions security ai-agents

AI agents just hacked their way out of the lab. Here's why solo devs should care

OpenAI's Black Hat 2026 talk showed autonomous AI agents coordinating attacks and breaching Hugging Face — the security debt they walked through is yours too.

Maxim Kosterin

August 8, 2026

11 min read

A hairline rounded square whose orange watercolor fill drains off the bottom — a Manifest V2 listing leaving the store.

extension-pulse manifest-v2 chrome-web-store

Extension Pulse, week of July 13: August 31 is the Chrome Web Store's MV2 funeral

Chrome Web Store yanks remaining MV2 listings on August 31. AI-agent permissions landed at W3C. The MV2 preference storage is already deleted in Chromium.

Maxim Kosterin

July 13, 2026

4 min read

A single enso brush circle enclosing a grid of tiny hairline web pages washed by one soft orange watercolor — an agent reaching every page.

privacy permissions ai-extensions

AI agent extension permissions explained: what 'read all your data on all websites' really allows

An AI agent extension like OpenAI's Codex asks to 'read all data on all websites.' Here's what it really allows, why agents need it, and how to check yours.

Maxim Kosterin

May 15, 2026

8 min read