Browser Devtools MCP is listed as active in the registry but did not answer our last check. 3 667 installs a week from npm.
Playwright browser & Node.js MCP. ARIA refs, screenshots, ref-based interactions.
The linked repository no longer exists on GitHub — it was deleted or made private.
Over the last week it answered 0.0% of our checks. We check every 15 minutes, so you hear about the next outage within the hour — not from your users.
Endpoint below is the one we actually reach during checks — not the one copied from a README. Last verified 12 min ago.
claude mcp add browser-devtools-mcp --transport http https://{your_mcp_server_host}/mcp
{
"mcpServers": {
"browser-devtools-mcp": {
"url": "https://{your_mcp_server_host}/mcp"
}
}
}
[mcp_servers.browser-devtools-mcp]
url = "https://{your_mcp_server_host}/mcp"
{
"mcpServers": {
"browser-devtools-mcp": {
"url": "https://{your_mcp_server_host}/mcp"
}
}
}
{
"mcpServers": {
"browser-devtools-mcp": {
"url": "https://{your_mcp_server_host}/mcp"
}
}
}
This one needs environment variables set before it will start:
PLATFORM (Platform: browser (default) or node for Node.js backend debugging.), PORT (Port for HTTP transport.), SESSION_IDLE_SECONDS (Idle session timeout (seconds).), SESSION_IDLE_CHECK_SECONDS (Interval for checking idle sessions (seconds).), SESSION_CLOSE_ON_SOCKET_CLOSE (Close session when socket closes.), CONSOLE_MESSAGES_BUFFER_SIZE (Maximum console messages to buffer.), HTTP_REQUESTS_BUFFER_SIZE (Maximum HTTP requests to buffer.), BROWSER_SERVER_INSTRUCTIONS_ENABLE (When true, include server instructions in MCP server info (browser platform). Default: true.), BROWSER_POLICY_UI_DEBUGGING_ENABLE (When true, include UI_DEBUGGING_POLICY in server policies (browser platform). Default: false.), BROWSER_HEADLESS_ENABLE (Run browser in headless mode.), BROWSER_PERSISTENT_ENABLE (Use persistent browser context (preserves cookies, localStorage, extensions).), BROWSER_CDP_ENDPOINT_URL (CDP attach: http(s) root or ws(s) URL. Resolves via /json/version, /json/list, or ws /devtools/browser (Chrome 136+). Mutually exclusive with BROWSER_PERSISTENT_ENABLE.), BROWSER_CDP_ENABLE (When true and no BROWSER_CDP_ENDPOINT_URL: probe 127.0.0.1:9222 then :9229 for CDP. Else use BROWSER_CDP_ENDPOINT_URL or default http://127.0.0.1:9222.), BROWSER_CDP_OPEN_INSPECT (On loopback CDP failure, if Chrome is running, open chrome://inspect/#remote-debugging. Default true. Set false to disable.), BROWSER_PERSISTENT_USER_DATA_DIR (Directory for persistent browser context user data.), BROWSER_USE_INSTALLED_ON_SYSTEM (Use system-installed Chrome browser instead of Playwright's bundled browser.), BROWSER_EXECUTABLE_PATH (Custom browser executable path.), BROWSER_CHROMIUM_SANDBOX (When false, launches Chromium with --no-sandbox. Required for Docker/Lambda/CI.), BROWSER_LAUNCH_ARGS (Extra Chromium launch args (comma-separated).), BROWSER_ALLOWED_DOMAINS (Comma-separated list of allowed hostnames. When set, all requests to non-matching hosts are aborted. Subdomains match automatically.), BROWSER_LOCALE (Browser locale (e.g., "en-US", "tr-TR").), OTEL_ENABLE (Enable OpenTelemetry integration.), OTEL_SERVICE_NAME (OpenTelemetry service name.), OTEL_SERVICE_VERSION (OpenTelemetry service version.), OTEL_ASSETS_DIR (Directory containing OpenTelemetry bundle files.), OTEL_INSTRUMENTATION_USER_INTERACTION_EVENTS (User interaction events to instrument (comma-separated, e.g., "click,submit").), OTEL_EXPORTER_TYPE (OpenTelemetry exporter type: "otlp/http", "console", or "none".), OTEL_EXPORTER_HTTP_URL (OpenTelemetry collector base URL (e.g., "http://localhost:4318").), OTEL_EXPORTER_HTTP_HEADERS (OpenTelemetry exporter HTTP headers (comma-separated key=value pairs).), AWS_REGION (AWS region for Bedrock usage.), AWS_PROFILE (AWS profile name for Bedrock usage.), AMAZON_BEDROCK_ENABLE (Enable Amazon Bedrock usage.), AMAZON_BEDROCK_IMAGE_EMBED_MODEL_ID (Amazon Bedrock image embedding model ID.), AMAZON_BEDROCK_TEXT_EMBED_MODEL_ID (Amazon Bedrock text embedding model ID.), AMAZON_BEDROCK_VISION_MODEL_ID (Amazon Bedrock vision model ID.), FIGMA_ACCESS_TOKEN (Figma API access token for design comparison.), FIGMA_API_BASE_URL (Figma API base URL.), DAEMON_PORT (Daemon mode port (CLI).), DAEMON_SESSION_IDLE_SECONDS (Daemon mode idle session timeout (seconds).), DAEMON_SESSION_IDLE_CHECK_SECONDS (Daemon mode idle session check interval (seconds).), NODE_SERVER_INSTRUCTIONS_ENABLE (When true, include server instructions in MCP server info (node platform). Default: true.), NODE_POLICY_DEBUGGING_ENABLE (When true, include NODE_DEBUGGING_POLICY in server policies (node platform). Default: false.), NODE_INSPECTOR_HOST (Inspector host for debug_connect when MCP runs in Docker (e.g. host.docker.internal). Default: 127.0.0.1.), TOOL_OUTPUT_SCHEMA_DISABLE (When true, omit tool output schema from MCP tool registration (can reduce token usage for some clients).), AVAILABLE_TOOL_DOMAINS (Comma-separated tool domains to enable (e.g. navigation,interaction,a11y,scenario). Unset = all tools. Browser: a11y, content, debug, figma, interaction, navigation, o11y, react, scenario, run, stub, sync. Node: debug, run.), WORKING_DIR (Working directory for project-level config (.browser-devtools-mcp/). Default: process.cwd().), SEARCH_STRATEGY (Search strategy: SIMPLE (default, minisearch + stemmer) or FTS5 (better-sqlite3, optional dep).), SCENARIO_SEARCH_STRATEGY (Override SEARCH_STRATEGY for scenario search only.).
The author declared them in the registry entry; get the values from the project itself.
| URL | Transport | State | Latency | Checked |
|---|---|---|---|---|
| https://{your_mcp_server_host}/mcp | streamable-http | answering | 0 ms | 12 min ago |
Drive a real Playwright browser: annotated screenshot + structured page state after every action.
Goal-driven Playwright browser sessions for LLMs, with full action traces and aria-snapshots.
Browser MCP: drive a real Chromium via Playwright with compact ref-addressable page snapshots
Fetch, parse, screenshot and browse pages via MCP without a real browser -- no JS execution.
Unity Editor MCP server: execute_code, screenshots, input simulation, play mode automation.
MCP server for AI agents to interact with terminal applications. Think Playwright for terminals.
Hosted Playwright API for AI agents. Returns a11y trees instead of screenshots.
Hosted browser for AI agents: screenshots, post-JS DOM, console, WCAG. No install, no API key.
Answers built from our own checks of this server.