browser-automation

# Browser Tool Use this skill after the compact `browser` tool prompt points you here. It is the progressive-disclosure workflow guide for rendered pages, multi-step browser work, logins, downloads, JavaScript-heavy sites, screenshots, host/container browser mode, and visual inspection. Prefer `search_engine` or `document_query` for plain text research. For fragile forms, load `browser-form-workflows` with `skills_tool:load` before acting when selects, checkboxes, radios, file uploads, contenteditable fields, validation, or final submission state are central to the task. ## Core Workflow 1. `open` creates a browser tab and returns a `browser_id`. 2. `content` returns readable markdown plus typed refs like `[link 3]`, `[button 6]`, `[input text 8]`. 3. Interact with refs using `click`, `type`, `submit`, `scroll`, etc.; iframe/shadow targets may return frame-chain metadata in action results. 4. Use `navigate` on an existing `browser_id` for serial browsing. 5. Keep only a small working tab set; close pages when finished. 6. If the user asks for an existing tab, page title, or already-open URL, call `list` first, match by `title` or `currentUrl`, then use `set_active` or `navigate` on that `browser_id` instead of opening a new tab. ## Modes When the user asks for "my browser", "host browser", "local browser", a local Chromium browser, or opening a URL in their host browser, use this `browser` tool. Do not substitute `computer_use_remote`, `code_execution_remote`, `xdg-open`, `sensible-browser`, or Python `webbrowser.open`. If setup fails and mentions remote debugging, tell the user to open the browser inspect page, such as `chrome://inspect/#remote-debugging` or `opera://inspect/#remote-debugging`, enable "Allow remote debugging for this browser instance", run `/browser host on`, and retry. The same tool may run in Docker container mode or A0 CLI host-browser mode, depending on project/plugin settings. - Container mode: browser and upload paths resolve inside the Agent Zero container. - Host mode: browser and upload paths resolve on the connected A0 CLI host machine. In host mode, page content and screenshots may be blocked by host-content policy when remote models are active. ## Screenshots And Vision Screenshots are explicit only; the browser does not automatically load images into model context. 1. Call `browser` with `action: "screenshot"`. 2. Call `vision_load` with the returned `vision_load.tool_args.paths` value. 3. Reason from the latest loaded screenshot. Screenshot args include `quality`, `full_page`, and optional `path`. Without `path`, the screenshot is saved as a chat-scoped artifact and returned through `vision_load.tool_args.paths`; with `path`, PNG is used when `path` ends with `.png`, otherwise JPEG is used. ## Forms And Files - `select_option` works for native selects and detectable ARIA listbox/combobox controls; pass one choice via `value` or several via `values`. - `set_checked` works for checkbox, radio, switch, and toggle-like refs; pass the desired state via `checked`. - `upload_file` works for file input `path` or multiple `paths`, file input refs or associated labels; verify files exist in the active browser environment. - For fragile forms, call `skills_tool` with `action: "load"` and `skill_name: "browser-form-workflows"`, then follow that form-specific workflow before filling or submitting. ## Pointer And Keyboard - Prefer refs/selectors and DOM/CDP actions over viewport coordinates. - Interaction targets use `ref` from the latest `content` output; `hover`, `double_click`, `right_click`, and `drag` also accept viewport coordinates when no reliable ref exists. - Coordinates are Chromium viewport CSS pixels and match screenshots; treat them as visual fallback, not the default interaction path. - Coordinate args: `x`/`y` position the pointer; `drag` adds destination `to_x`/`to_y` or destination `target_ref`; element-relative offsets use `offset_x`/`offset_y` and drag `target_offset_x`/`target_offset_y`. - `mouse` sends raw pointer input with `event_type` (default `click`) plus `x`/`y` and `button`. - `click` with `modifiers` (e.g. `"Control"`) alters the click; `focus_popup: true` targets an auto-registered popup window. - `key_chord` presses `keys` in order and releases in reverse; `keyboard` types `text` or presses a single `key`. - `content` narrows extraction via `selector` or a list of `selectors`. - `evaluate` runs JavaScript in the page via `script` and returns the evaluated result. - Browser settings set the evaluate deadline (default 30 seconds, range 0.1–60 seconds); there is no tool-call timeout override. A timeout interrupts JavaScript in place and preserves the page when execution stops. If async execution remains pending, recovery may reload the affected tab, losing unsaved DOM edits; cookies and tab storage normally survive. The error explicitly reports a reload or fallback closure; use `list` before retrying a closed tab. Host evaluation requires an updated connector; Safari rejects it because its backend cannot forcibly interrupt JavaScript. - `wheel` scrolls by pixel deltas `delta_x`/`delta_y` at `x`/`y`. - `clipboard` actions are copy, cut, or paste, chosen via `clipboard_action`; paste inserts `text`. - `set_viewport` resizes the page viewport via `width` and `height`. - `navigate` takes the destination `url` on an existing `browser_id`. ## Tabs And Popups - Popups and target-blank tabs are auto-registered. - `list` shows open tabs; pass `include_content: true` sparingly. - `set_active` deliberately changes focus. - Operations on a non-active tab do not steal focus unless browser rules require it. ## Browser Action Multi `multi` is only a browser action, never a top-level tool. Use: ```json { "tool_name": "browser", "tool_args": { "action": "multi", "calls": [ {"action": "content", "browser_id": 1}, {"action": "screenshot", "browser_id": 2} ] } } ``` Use browser action `multi` for parallel reads across tabs. Avoid mutating the same tab twice in one batch unless serial order is intended.