Agentic Hat ("the Extension") lets you collect images from websites you visit and send them into your Agentic Hat agent conversation. It can also, at the request of a conversation you are signed in to and only on websites you approve, read and interact with web pages on your behalf ("browser tools"). This policy explains exactly what the Extension accesses, how it is used, and where it is stored.
The Extension only acts on images you explicitly choose. When you click the collect button on an image, it accesses the image itself (its URL and a copy of its image data) and the URL of the page you collected it from (kept as a source reference).
Outside of browser tools (below), it does not read, record, or transmit the general content of the pages you browse, your browsing history, form inputs, keystrokes, or any personal account information.
When you ask your Agentic Hat conversation to use the browser, that conversation may ask the Extension to list your open tabs (domain names only, never full addresses), open a page, navigate or close a tab it opened itself, bring a tab to the front, read a page’s title and main text, extract its images, read the labels and state of its interactive elements, find an element by its text, scroll, hover, press a key, fill in a field, click an element, place an image into a file-upload field, or capture a screenshot of the visible part of the tab.
Such requests can only come from an Agentic Hat tab you are signed in to; no other website or service can send them. Before the first read of any website, and again before the first interaction with it, the Agentic Hat app asks for your permission. Nothing happens on a site you have not approved, and browser-internal pages are never accessible.
Approvals are stored per website in your browser’s extension storage (synced with your Chrome profile if Chrome sync is on) and can be reviewed and revoked at any time under “Sites your chat may use” in the side panel settings (uninstalling the Extension also clears them). Tabs the conversation opens or acts on are placed in a “👀 Buddy” tab group so you can see and close them together.
Page text, images and screenshots obtained this way are handed to your Agentic Hat conversation through your own signed-in tab and are then processed by the Agentic Hat service under https://agenti.chat/privacy. The Extension itself keeps no copy of page text.
CAPTCHA and security-check pages. The Extension will not solve, click, or submit a CAPTCHA for you. To recognise such pages it uses a small list of page patterns bundled with the Extension, refreshed about once a day by downloading a static file from our own CDN (https://cdn.agenti.chat); this is a download only — nothing about you or your browsing is sent with it, and no code is executed from it. When a requested page is recognised as a CAPTCHA, the Extension declines the request and tells your conversation the shape of the page that was blocked — its domain, its path with numeric or token-like segments removed, the names (not values) of its query parameters, and the domains of any embedded verification frames — so the agent can ask you to complete the check yourself. Full page addresses and page content are not sent. If you mark a page as a CAPTCHA yourself in the side panel, that page shape is stored in your browser’s extension storage (synced with your Chrome profile if Chrome sync is on), is used only on your own browser, and can be reviewed and deleted under “Pages you marked as CAPTCHA” in the side panel settings.
Collected images are used solely to let you review them and send them into your Agentic Hat conversation as attachments for AI image tasks. Page content obtained by browser tools is used solely to fulfil the request you made in that conversation. The Extension has no other function.
Collected images and their source links are stored locally on your device, in the browser’s IndexedDB / extension storage.
The Extension does not upload your collected images to any FewBox server of its own, and does not operate any independent tracking, analytics, or advertising.
When you (or the automatic send of the latest image) push a collected image into an open Agentic Hat web app tab, or when a browser tool returns page content to your conversation, that data is handed to the Agentic Hat web application under your own signed-in account. From that point the image is processed by the Agentic Hat service to fulfill your request and is governed by the Agentic Hat Privacy Policy at https://agenti.chat/privacy. The Extension itself does not send your data anywhere else.
• We do not sell or rent your data.
• We do not use it for advertising or profiling.
• We do not track your browsing across sites.
• We do not collect personally identifiable information through the Extension.
• Access to all websites — to show the collect button on any site and to fetch the bytes of the specific image you choose to collect; and, for browser tools, to fetch images from and run the actions above in websites you have approved.
• tabs — to locate your open Agentic Hat app tab so a collected image can be delivered to it; and, for browser tools, to list open tabs (domain only), open a page, and wait for it to load.
• tabGroups — to place tabs the conversation opens or acts on in a named “👀 Buddy” group.
• scripting — to re-attach to already-open Agentic Hat tabs after an update, and to run browser-tool actions inside websites you have approved.
• storage / unlimitedStorage — to keep your collected images locally on your device.
• sidePanel — to display the collection panel.
Collected images remain on your device until you remove them (delete or clear them in the side panel) or uninstall the Extension, which removes all locally stored data, including per-website approvals for browser tools.
The Extension is not directed to children under 13 and does not knowingly collect data from them.
We may update this policy; material changes will be reflected by a new effective date on this page.
Questions: [email protected]