MCP Server

What VisualSentinel/mcp-server does

VisualSentinel/mcp-server MCP server exposes Visual Sentinel's monitoring API through the Model Context Protocol. Its coverage spans uptime, performance, SSL certificates, DNS, visual regression, and content changes, although the documented tool list focuses on public diagnostics and account-level monitoring data.

The server provides 17 tools. Five are available without authentication: service health, DNS record lookup, TLS certificate inspection, a one-off speed test, and a basic website check. The remaining tools can list, inspect, create, and check monitors; retrieve uptime and bulk metrics; inspect incidents; list or acknowledge alerts; list public status pages; and list registered origin servers.

It also includes four prompts for website health checks, broken-site investigation, pre-launch review, and open-incident triage. Three read-only resources describe plans, setup help, and the available tools.

How it works

The package runs as a standard MCP process and communicates with the client using JSON-RPC over standard input and output. An MCP client can select tools based on the user's request and return the Visual Sentinel API response. The public checks can query a site directly without an account.

Authenticated operations use the VS_API_KEY environment variable. These operations include monitor management, incident and alert access, status-page listing, and server listing. If no key is configured, the process still starts and advertises its tools; an authenticated call returns an explanation that a key is needed rather than silently failing.

Some operations change state. Creating a monitor, triggering an immediate monitor check, and acknowledging an alert use the permissions attached to the API key. A read-only key cannot create monitors.

Setup and configuration

Install the published npm package as @visualsentinel/mcp-server, or run it through npx -y @visualsentinel/mcp-server. The README documents configurations for Claude Desktop, Claude Code, Cursor, Cline, Continue, Docker, Gemini CLI, and generic MCP clients. It also describes repository plugin files for Cursor and Claude Code-related use.

Set VS_API_KEY when account tools are needed. API keys are available to Visual Sentinel Business and Agency plans. VISUALSENTINEL_API_KEY is accepted as an alternative variable name. VS_API_BASE defaults to https://visualsentinel.com and is intended for testing against a local development instance.

For Claude Desktop, add the package as an npx command in the MCP server configuration and restart the client. Docker users build the included image and run it interactively with the key passed through the environment.

Tools and capabilities

VisualSentinel/mcp-server MCP server supports these main workflows:

• Resolve A, AAAA, MX, NS, TXT, CNAME, and SOA records.
• Inspect certificate issuer, validity, SAN entries, and certificate chain.
• Measure time to first byte, total load time, and transfer size.
• Check HTTP website health and Visual Sentinel service health.
• Filter monitors, incidents, and alerts by the documented status or monitor fields.
• Retrieve uptime for preset windows, custom ranges, or full history.
• Request uptime or response-time metrics for up to 100 monitors in one call.
• Create monitors, run immediate checks, and acknowledge alerts.

Limitations and notes

VisualSentinel/mcp-server MCP server depends on the Visual Sentinel API for its monitoring data and authenticated operations. The account tools require a Business or Agency plan with an API key; the five public tools do not. Public-tool results may appear in conversation logs, so access to those logs should be treated as access to the returned site information.

The API key is kept by the machine running the MCP process and requests are sent to Visual Sentinel over HTTPS. Revoke a compromised key from Visual Sentinel's API-key settings. The documented server transport is stdio, and the tool surface follows the public Visual Sentinel API specification.