zen
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
No arguments | |||
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": true
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| tabs_listA | List open tabs in Zen as "[tabId] title — url", with the tab the user is looking at first. Call this first. Shows up to |
| tab_createA | Open a new tab in Zen, optionally at a URL, and wait for it to load. It opens in the same container as the page the user is on, so it shares the same logins. Prefer working in a tab you created over reusing one of the user's tabs. |
| tab_closeC | Close a tab. |
| navigateB | Navigate a tab to a URL, or go "back", "forward" or "reload". Waits for the page to finish loading. |
| read_pageA | Get a structured outline of the page: elements with their role, name and a ref ID to use with click/type/form_input. filter "interactive" (default) lists only buttons, links and inputs; "all" also includes headings, text and images. |
| get_page_textB | Get the readable text of the page (article/main content when present). |
| findA | Find elements whose text, label, placeholder, role or CSS selector matches a query. Returns refs. Plain text match, not natural language: use words that appear on the element ("Sign in", "search"), or a CSS selector. |
| clickC | Click an element by ref or by viewport coordinate. |
| hoverA | Move the mouse over an element (by ref or coordinate) to reveal menus and tooltips. |
| typeA | Type text into an element (by ref), or into whatever is focused. Inserts at the cursor; use form_input to replace a field's whole value. |
| keyB | Press a key or chord on the focused element, e.g. "Enter", "Escape", "Tab", "ArrowDown", "Backspace", "ctrl+a". |
| form_inputB | Set the value of a form control by ref: text fields (replaces the value), checkboxes/radios (true/false), and selects (option value or visible label). |
| scrollA | Scroll the page or a scrollable area. With ref and no direction, scrolls that element into view. |
| annotateA | Point things out on the page itself: draws an outline around elements (by ref) or highlights a passage (by quoting its text), each with an optional short label. Use this to show the user where something is instead of describing its position. Replaces earlier marks unless keep is true. |
| clear_annotationsB | Remove all marks drawn on the page by annotate. |
| get_selectionA | Get the text the user currently has selected on the page, with the surrounding passage for context. |
| show_cardA | Show a floating card on the page containing HTML you write: a diagram, an animation, a worked example, a small interactive demo. The HTML must be self-contained (inline , , SVG or canvas); it runs isolated with no network access, so no external scripts, fonts, images or fetch (data: images are fine). Design it for the given size without scrolling, and support light and dark with prefers-color-scheme. The user can drag and close the card. A new card replaces the ones you showed before unless keep is true. The result says whether the content actually drew; if it reports an error or an empty body, fix the HTML rather than telling the user it is there. |
| close_cardsB | Close the cards you put on the page with show_card. |
| screenshotA | Take a screenshot of the tab's visible viewport. Pixel positions in the image are the coordinates click/hover/scroll accept. |
| javascriptA | Run JavaScript in the page and return the value of the last expression (a promise is awaited). Top-level |
| wait_forB | Wait until text or a CSS selector appears on the page, or just for a number of seconds. |
| console_messagesA | Read console output and uncaught errors captured from a tab since it was opened or last cleared. |
| network_requestsA | Read the network requests a tab has made (method, URL, status, type) since it was opened or last cleared. |
| resize_windowC | Resize the browser window containing a tab. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 24 tools
Most tools target a clearly distinct action (click, hover, type, form_input, key) and the descriptions explicitly separate the close pairs (type vs. form_input vs. key; read_page vs. get_page_text vs. find). A few boundaries are softer — navigate vs. tab_create-at-URL, and annotate vs. show_card both draw things on the page — but the descriptions resolve these well enough that misselection is unlikely.
Two conventions coexist: around half the tools use verb_noun snake_case (form_input, clear_annotations, read_page, show_card, wait_for, resize_window) while others are bare single-word verbs (click, type, key, scroll, find, hover, navigate, screenshot, javascript). The tab tools also invert the order to noun_verb (tabs_list, tab_create, tab_close) instead of list_tabs/create_tab. It stays readable, but the pattern is genuinely mixed.
24 tools sits in the heavy-but-defensible band: browser automation legitimately needs navigation, tab management, page reading, input, and debugging surfaces. Still, several are niche (resize_window, close_cards, clear_annotations, network_requests) and push the set past the point where an agent can hold the whole surface in mind cheaply.
Coverage of the browsing lifecycle is strong: tab create/list/close, navigation, structured and text reading, search, input, screenshots, JS execution, console and network inspection, plus waiting and window sizing. Minor gaps remain — no explicit tab activation/focus switch for a specific existing tab, no drag-and-drop, no cookie/storage or download handling — though most of these can be reached through javascript.