OpenBrowser
An enhanced MCP Playwright browser server that enables robust web automation with persistent locator caching, multi-level click fallbacks, accessibility tree interactions, and smart handling of navigation and popups.
Caution. Specific findings reduced this grade. They are listed on the page. Grades reflect the full trust pyramid: code, verification depth, and reputation. New projects cap at C until adoption is earned.
How we verified
Verified is a snapshot. Live keeps it current, and builds your track record.
⚡ Connect GitHub → continuous verification on every pushwhy connect →Who stands behind it
SangoAlgo
Source: Glama
Claim it to get a verified publisher badge, a free copy of our full audit findings, and direct contact for any high-priority issues we find. Or connect your repo for our deepest verification, Live Monitored: read-only, revoke anytime. What we access →
Install from
The grade above is for the source repository. Registries can serve a different version, so we mark the ones we were not able to read.
These names and descriptions are the publisher's own, read from the source code. We print them as written. Our assessment is the findings above, not this list.
browser_get_accessibility_treeAccessibility tree snapshot. Accepts scope parameter to filter by region (e.g. scope='left-panel', scope='navigation'). Returns compact tree with aN ref IDs. Ref IDs are cacheable — can be used in browser_click_a11y for elements invisible to DOM (portals, overlays).
browser_click_a11yClick element by accessibility-tree reference (e.g. 'a3' from browser_get_accessibility_tree). Uses Playwright's accessibility snapshot API — works even for elements invisible to DOM queries (React portals, shadow DOM, off-screen but a11y-visible). Falls back to role+name locator if ref cache is sta…
browser_clickClick element by CSS, XPath, text, role=role[name='...'], or ref (from compact map). Auto-fallback: Playwright click -> JS click on visibility/overlay errors -> a11y role+name search. Accepts ref='e5' from compact_map — uses cached stable locator (data-tsid, aria-label, role+name).
browser_force_clickForce-click element via JS — bypasses visibility, overlays, and pointer-event interception. Accepts same selector types as browser_click.
browser_fillFill input field or contenteditable div. Auto-detects element type: standard inputs use Playwright fill(), contenteditable divs use execCommand('insertText') + input event dispatch for web-component compatibility.
browser_get_compact_mapCompact element map with persistent ref IDs [e1], [e2]... (60 max, paginated). Ref IDs are stable within a tab session — cached by data-tsid > aria-label > role+name. Accept optional scope selector to filter region. NOW INCLUDES a11y-only elements — items visible to screen readers but hidden from CS…
browser_get_deltaDOM Mutation Delta — only changed elements since last snapshot (~5 tokens). Also reports URL changes and popup openings.
browser_get_contentGet page content in text, HTML, or innerText format. Optional CSS selector to scope.
browser_evaluate_jsExecute JavaScript in page context. Returns JSON-serialized result.
browser_take_screenshotTake full-page or element screenshot.
browser_navigateNavigate to URL. Auto-waits for DOM stability (network idle + mutation settle). Returns compact action targets. Automatically detects and captures popup windows (OAuth flows). No manual sleep needed — built-in stabilization.
browser_go_backNavigate back in history with smart-wait.
browser_list_tabsList all open tabs (including popups opened via OAuth). Returns tab IDs and URLs.
browser_switch_tabSwitch active focus to a specific tab. Auto-activates popup tabs for OAuth flows.
browser_smart_waitSmart wait — waits for DOM stability (network idle + no mutations for 500ms). Use instead of manual sleep(). Auto-tracks mutation observers for detect SPA re-renders.
browser_wait_for_navigation_or_popupWait for any outcome after an action: URL change, popup open, or network idle. AUTO-SWITCHES to popup tab if detected. Returns outcome type + new URL/tab_id. Use after clicks that may trigger OAuth popups or redirects. Eliminates manual sleep(3000-5000).
Disclosed vulnerabilities in this server's declared npm dependencies (via OSV). Whether each is reachable depends on the installed versions.
Anthropic's MCP TypeScript SDK has a ReDoS vulnerability
Model Context Protocol (MCP) TypeScript SDK does not enable DNS rebinding protection by default
Playwright downloads and installs browsers without verifying the authenticity of the SSL certificate
Tool annotations
No tools have read-only/destructive annotations
Add readOnlyHint or destructiveHint annotations to every tool so hosts can warn users before invoking.
All four hints declared on every tool
16/16 tools missing one or more hints — browser_get_accessibility_tree (missing: readOnlyHint, destructiveHint, idempotentHint, openWorldHint); browser_click_a11y (missing: readOnlyHint, destructiveHint, idempotentHint, openWorldHint); browser_click (missing: readOnlyHint, destructiveHint, idempotentHint, openWorldHint), +13 more. OpenAI's directory rejects tools where any of the four hints are missing or non-boolean.
For every tool, set all four hints (readOnlyHint, destructiveHint, idempotentHint, openWorldHint) to explicit true/false values that match the handler’s actual behaviour.
License file
No license file
Add a LICENSE file (MIT, Apache-2.0, etc.).
Tests exist
No test files found
Add tests that exercise each declared tool.
Production dependencies are patched
0 critical, 3 high severity in production deps — @modelcontextprotocol/sdk@1.6.1 (high), @modelcontextprotocol/sdk@1.6.1 (high)
Run npm audit fix, or upgrade the affected packages to a non-vulnerable version.
Claim the listing to review these findings one by one and send us a correction where you disagree, straight to the team. Claiming also means we tell you when the grade moves, and reach you first if we find anything urgent.
[](https://m8ven.ai/mcp/sangoalgo/openbrowser)?variant=verified from the URL.Vetting this one by hand? Tool Check is an MCP that scores other MCPs. Add it once and ask Claude, ChatGPT, or any MCP client to grade a server, surface CVEs, check the publisher, and suggest safer alternatives — before you install.
https://m8ven.ai/api/mcp/tool-check