devtools-fleet-mcp
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| DEVTOOLS_FLEET_HOME | No | Moves the whole ~/.devtools-fleet directory to another location. | |
| DEVTOOLS_FLEET_CHANNEL | No | Chrome release channel: stable, beta, dev or canary. Default: stable. | stable |
| DEVTOOLS_FLEET_SESSION | No | Replaces the process-tree session lookup with a fixed label, useful in CI or containers. | |
| DEVTOOLS_FLEET_HEADLESS | No | Whether to launch browsers headless. Default: true. | true |
| DEVTOOLS_FLEET_VIEWPORT | No | Default viewport, e.g. "1280x720". No default. | |
| DEVTOOLS_FLEET_CHROME_PATH | No | Path to a Chromium/Chrome build. Default: detected. | |
| DEVTOOLS_FLEET_MAX_BROWSERS | No | Maximum number of browsers across the machine. Default: 10. | 10 |
| DEVTOOLS_FLEET_LAUNCH_TIMEOUT_SECONDS | No | Launch timeout in seconds. Default: 30. | 30 |
| DEVTOOLS_FLEET_ORPHAN_TIMEOUT_MINUTES | No | How long a dropped connection's browser waits for a reconnect, in minutes. Default: 15. | 15 |
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 | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| browser_startA | Start this session's own Chrome (isolated from every other agent). Optional: the first browser tool call starts one with defaults anyway. Call it explicitly to load a saved login ( |
| browser_statusA | Show this session's browser: id, saved state, allowlist, headless, open tabs, uptime. |
| browser_restartA | Relaunch this session's browser keeping cookies, localStorage and open tabs. Pass headless: false to get a visible window (e.g. so the human can watch or solve a captcha), headless: true to hide it again. Pages reload. |
| browser_stopA | Close this session's browser and delete its temporary profile. |
| state_saveA | Save the current browser's login (cookies + localStorage) as a named state, so other sessions can start already logged in. Only cookies for the allowed origins are kept. allowedOrigins defaults to the http(s) origins open in tabs; it can never be wider than this browser's own allowlist. States a person created with |
| state_listA | List saved login states: name, allowed origins, cookie counts, when saved and when the first cookie expires. Never shows cookie values. |
| clickC | Clicks on the provided element |
| close_pageA | Closes the page by its index. The last open page cannot be closed. |
| dragC | Drag an element onto another element |
| emulateC | Emulates various features on the target page. |
| evaluate_scriptB | Evaluate a JavaScript function inside the target page. Returns the response as JSON, so returned values have to be JSON-serializable. |
| fillC | Type text into an input, text area or select an option from a element. |
| fill_formA | Fill out multiple form elements (inputs, selects, checkboxes, radios) at once. ALWAYS prefer this tool over multiple individual 'fill' or 'click' calls when interacting with forms. It is significantly faster, more reliable, and reduces turn count. Example: Fill username, password, and check "Remember Me" in one call. |
| get_console_messageA | Gets a console message by its ID. You can get all messages by calling list_console_messages. |
| get_css_stylesA | Retrieve matched CSS rules, inline styles, inherited styles, and cascade information for an element identified by its UID. Use this tool to debug why specific CSS properties are applied, overridden, or conflicting. Results are paginated and return 10 rules per page by default; use pageIdx to page through the remaining rules. Requires a UID from take_snapshot. |
| get_network_requestB | Gets a network request by an optional reqid, if omitted returns the currently selected request in the DevTools Network panel. Useful for inspecting request headers (including 'Cookie') and response headers (including 'Set-Cookie' and directives). |
| handle_dialogC | If a browser dialog was opened, use this command to handle it |
| hoverC | Hover over the provided element |
| lighthouse_auditA | Get Lighthouse score and reports for accessibility, SEO, best practices, and agentic browsing. This excludes performance. For performance audits, run performance_start_trace |
| list_console_messagesB | List all console messages for the target page since the last navigation. |
| list_network_requestsB | Lists the most recent requests for the target page since the last navigation. |
| list_pagesB | Get a list of pages open in the browser. |
| navigate_pageB | Go to a URL, or back, forward, or reload. Use project URL if not specified otherwise. |
| new_pageB | Open a new tab and load a URL. Use project URL if not specified otherwise. |
| performance_analyze_insightB | Provides more detailed information on a specific Performance Insight of an insight set that was highlighted in the results of a trace recording. |
| performance_start_traceA | Start a performance trace on the target webpage. Use to find frontend performance issues, Core Web Vitals (LCP, INP, CLS), and improve page load speed. |
| performance_stop_traceB | Stop the active performance trace recording on the target webpage. |
| press_keyA | Press a key or key combination. Use this when other input methods like fill() cannot be used (e.g., keyboard shortcuts, navigation keys, or special key combinations). |
| resize_pageB | Resizes the page's window so that the page has specified dimension |
| select_pageB | Select a page as a context for future tool calls. |
| take_heapsnapshotA | Capture a heap snapshot of the target page. Use to analyze the memory distribution of JavaScript objects and debug memory leaks. |
| take_screenshotC | Take a screenshot of the page or element. |
| take_snapshotA | Take a text snapshot of the target page based on the a11y tree. The snapshot lists page elements along with a unique identifier (uid). Always use the latest snapshot. Prefer taking a snapshot over taking a screenshot. The snapshot indicates the element selected in the DevTools Elements panel (if any). |
| type_textA | Type text using keyboard into a previously focused input |
| upload_fileC | Upload a file through a provided element. |
| wait_forC | Wait for the specified text to appear on the selected page. |
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 36 tools
Most tools target distinct resources or actions, such as take_snapshot vs take_screenshot and list_console_messages vs get_console_message. However, the input family (click, fill, fill_form, type_text, press_key) has some overlapping boundaries, though descriptions clarify preferred usage.
The set is predominantly snake_case and verb_noun (close_page, list_pages, take_snapshot). A minority use noun-first or noun-verb ordering (browser_status, state_save, performance_start_trace), which is a minor deviation but still readable.
36 tools is high, but the domain (full browser automation and DevTools) legitimately requires separate tools for lifecycle, pages, input, inspection, performance, network, console, and state. It is slightly over the typical 3-15 range but each tool earns its place.
The surface covers browser lifecycle, navigation, page/tab management, input, a11y snapshots, screenshots, console, network, CSS, performance tracing, Lighthouse, memory heap, dialogs, file upload, emulation, and login states. No obvious lifecycle gaps for a DevTools automation server.