The hosted Visual Sentinel MCP server adds 106 monitoring tools to any AI client that speaks Model Context Protocol. Ask Claude what is down, triage an incident, change a monitor's settings, or create a new visual-regression monitor without leaving the conversation.
MIT · ESM · Node 18+ · 5 tools work without an API key
hosted server · nothing to install
Connect from Claude, ChatGPT and other assistants
Add Visual Sentinel as a remote MCP server (streamable HTTP). Your assistant asks you to sign in once, then calls all 106 tools. The local server below runs the 17 core ones. Setup for each assistant is in the next section.
If Claude itself stops answering, see Claude status for our latest check of claude.ai and the Claude API.
Server URL
https://visualsentinel.com/api/mcp
Sign-in
OAuth. You sign in to Visual Sentinel, pick a workspace if you have more than one, and approve what the assistant asks to do. There is no key to copy.
Read access
See your monitors, incidents, alerts, alert channels, maintenance, status pages, servers and workspace settings
Write access
Create, change and delete monitors, alert channels and rules, incidents, maintenance, status pages, servers and workspace settings
Plan
The 101 account tools work on every paid plan, like API keys. On other plans the assistant can still connect and use the 5 public tools.
Without an account
https://visualsentinel.com/api/mcp/public serves the 5 public tools with no sign-in.
Limits
Without an account the free tools allow 300 calls a day and about 10 a minute per IP address (not per domain checked). A connected account gets 120 calls a minute per connection, and the free check tools stay at about 10 a minute within that.
Disconnect
Remove an assistant at any time in Workspace settings, under Developer. It loses access on its next request.
connect your AI agent
Setup for each assistant
Signed in to Visual Sentinel? Open Connect your AI agent at the top of any workspace page for the same steps with copy buttons. Or hand your assistant visualsentinel.com/agents/setup.md, instructions it can follow from start to finish.
Adds Visual Sentinel to Cascade in Windsurf, now Devin Desktop, with an API key.
Create an API key in Visual Sentinel: Workspace settings, then Developer access. Pick read only, or read and write if the agent should make changes.
In the Cascade panel, open the ... menu at the top right and click Open MCP config file in the MCPs section. On macOS and Linux it is ~/.config/devin/mcp_config.json, on Windows %APPDATA%\devin\mcp_config.json.
Add the entry below and save.
Set VS_API_KEY to the key in the environment the app starts from. Keep the key out of chats and out of files you share.
Replace the placeholder with your API key from /settings/api-keys. Restart your client; tools become available immediately.
Claude Desktop
json
Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows). Restart Claude Desktop afterwards.
Install the marketplace plugin (recommended, includes 3 skills and 3 slash commands) or just register the MCP server directly.
# Recommended: full plugin (MCP server + skills + commands)
claude plugin marketplace add github.com/VisualSentinel/claude-plugins
claude plugin install visual-sentinel@visualsentinel
export VS_API_KEY=vs_live_xxxxxxxxxxxxxxxxxxxxxxxxxxxx
# Or: MCP server only
claude mcp add visualsentinel -- npx -y @visualsentinel/mcp-server
Cline (VS Code extension)
json
Open Cline’s settings panel, MCP servers tab, and add the JSON below.
5 public tools work on any URL or domain without authentication. 101 authenticated tools manage the website monitors, their settings, alert channels and rules, incidents, maintenance, status pages, workspace settings and servers in your organization. The lists below are the 17 core tools, which the npm package runs too.
Public · no API key
vs_health
Check whether Visual Sentinel itself is up
vs_dns_check
Resolve A, AAAA, MX, NS, TXT, CNAME, or SOA records for a domain
vs_ssl_check
Inspect a TLS certificate (issuer, expiry, SAN list, chain)
vs_speed_test
Lighthouse lab test: performance score, LCP, FCP, CLS, TTFB, Total Blocking Time and Speed Index
vs_website_check
Quick HTTP health check on any URL
Authenticated · VS_API_KEY
vs_monitors_list
List monitors filtered by status or type
vs_monitors_get
Fetch one monitor with current status and last check
vs_monitors_create
Create a new HTTP, HTTPS, KEYWORD, PING, PORT, DNS, SSL, VISUAL, CONTENT, API, GRAPHQL, or WEBSOCKET monitor
vs_monitors_check_now
Trigger an immediate check outside the scheduled cadence
vs_monitors_uptime
Uptime percentage and outage breakdown over 24h, 7d, 30d, 90d, 365d
vs_monitors_metrics_bulk
Uptime or response-time metrics for up to 100 monitors in one request, over a preset window, a custom date range or the full history
vs_incidents_list
List incidents filtered by status or monitor
vs_incidents_get
Fetch one incident with full check history and root-cause hints
vs_alerts_list
List alerts by status
vs_alerts_acknowledge
Acknowledge an alert with an optional note
vs_status_pages_list
List your public status pages
vs_servers_list
List registered origin servers
what you can ask
Talk to your monitors, get real answers.
Status check at a glance
"What Visual Sentinel monitors are down right now?" Claude calls vs_monitors_list filtered by status, then summarises with the affected monitors and their last successful check.
Triage an incident without leaving the terminal
"Triage incident clm_xyz." Claude pulls the incident, scans the check history, and proposes a probable cause and a next action. Acknowledge directly with a follow-up message.
Onboard a new monitor by description
"Watch https://shop.example.com/checkout for visual changes." Claude checks the plan capability, proposes a valid visual cadence, confirms the monitor name with you, and creates it with vs_monitors_create.
Free SSL or DNS check on any domain
"What is the SSL situation on example.com?" Claude calls vs_ssl_check (no API key required) and surfaces issuer, expiry, SAN list, and chain. Same for DNS records, page speed, or a quick HTTP probe.
where your data goes
Local server. Local key. Direct HTTPS to us.
The MCP server runs locally on your machine via npx. Your VS_API_KEY never leaves your computer except in the direct HTTPS request to visualsentinel.com.
The AI client (Claude, Cline, Cursor, Continue) sees the JSON results of each tool call, just like it would any other tool output. Treat tool output the same way you treat any text in your conversation.
Public tools (DNS, SSL, speed, website check, health) use no key at all. Their results are still visible in your conversation.
Revoke a leaked API key instantly at /settings/api-keys. Revocation takes effect within seconds.
A Model Context Protocol (MCP) server for Visual Sentinel. It is hosted at https://visualsentinel.com/api/mcp, so there is nothing to install, and has 106 tools over the Visual Sentinel REST API: an AI assistant like Claude can list your monitors, create new ones, change their settings, set up alert channels, status pages and maintenance, fetch incident history, run free DNS or SSL or speed checks, and acknowledge alerts on your behalf. The npm package is a small Node.js program that runs the 17 core tools on your own machine.
Which AI clients does it work with?
Any client that can add a remote MCP server over streamable HTTP. Claude Code, Claude (web and desktop), ChatGPT, Cursor, Codex, VS Code, Antigravity, Gemini CLI and Windsurf have setup steps on this page. Most sign in with OAuth: your assistant opens Visual Sentinel in the browser and you choose the workspace and its access. Codex, Gemini CLI and Windsurf use an API key sent as a bearer token. The npm package still runs locally over stdio for clients that only support local servers.
Do I need a Visual Sentinel account?
For the public tools (DNS lookup, SSL inspection, speed test, website check, service health), no. Those work without authentication. For monitor management, incident triage, and alert acknowledgement you need a Visual Sentinel account and an API key generated at https://visualsentinel.com/settings/api-keys. New workspaces get a free first check of their site, then a 7-day trial of a paid plan (card required) or a subscription; API keys come with every paid plan.
How do I install it?
Pick your assistant in the Connect section of this page: Cursor and VS Code install from a link, Claude Code, Codex and Gemini CLI take one command or a few lines of settings, and Claude and ChatGPT take a few steps in their settings. Signed-in users can also open Connect your AI agent at the top of any workspace page. Or hand your assistant https://visualsentinel.com/agents/setup.md, which it can follow from start to finish.
Where does my API key go? Is it sent anywhere I do not control?
Store VS_API_KEY in your local MCP client configuration. The local server reads it from the environment and sends it only to visualsentinel.com over HTTPS for authenticated requests; it does not return the key in tool results. Anyone with access to your machine or configuration may be able to read it, so revoke a leaked key at https://visualsentinel.com/settings/api-keys.
Is there a Claude Code plugin too?
Yes. The visualsentinel marketplace bundles the MCP server with three task-focused skills (monitor status, incident triage, monitor onboarding) and three slash commands (/vs-status, /vs-incidents, /vs-add-monitor). Install with: claude plugin marketplace add github.com/VisualSentinel/claude-plugins, then claude plugin install visual-sentinel@visualsentinel.
What is Model Context Protocol?
Model Context Protocol (MCP) is an open standard for connecting AI assistants to tools and data sources. An MCP client (like Claude Desktop) speaks JSON-RPC to MCP servers (like this one) over stdio or HTTP. Each server advertises a set of tools the client can call. MCP is supported by Claude, Cline, Cursor, Continue, and a growing list of other AI products.
trusted by operators
What customers say
Real reviews from teams already running Visual Sentinel monitoring against their production sites.
G2 verified review
Instant alerts that make website monitoring stress-free
I used to manually check every client website to make sure it was still online. If a plugin update, SSL issue, or hosting problem caused downtime, I usually found out only after a client complained. Visual Sentinel fixed that for me. It monitors all my sites and sends instant alerts when something goes wrong, whether it is downtime, performance degradation, certificate issues, DNS changes, or visual breakage. The support for multiple alert channels, including WhatsApp, makes it even more useful. It has saved me a lot of time and made my website management far less stressful.
Visual Sentinel picked up and detected a rogue XMR crypto miner that injected itself via Magento. Alerted me of the issues, made it obvious what the problem was, and I could get repairs done immediately. Took us 18 minutes from start to finish to completely scrape out this garbage after diagnosis finished, and around an hour of downtime total. Wouldn’t have had any idea without your tools.
six layers · in your AI client
5 tools need no account. Add monitors when you are ready.
The public tools work without an account. Monitor management uses an API key, included with every paid plan. Start a 7-day trial (card required) or subscribe now.