Model Context Protocol · Hosted

    Plug Lemwatch into your AI assistant

    A best-in-market MCP server for website monitoring. Tools, live resources, and slash-command playbooks — with proper RBAC and confirmation gates on every write.

    19 tools, 5 resources, 6 playbooks

    More than any competitor's MCP. Read sites, incidents, SSL, SEO; write actions all gated by confirmation.

    Org-scoped, RBAC-aware

    Tokens inherit your organization role. Admin-only tools hidden from members. Free-tier clients still work.

    Hosted & instant

    No local install. Streamable HTTP transport. Works with Claude Desktop, Cursor, ChatGPT, and any MCP client.

    1. Mint a token

    Visit Settings → MCP Tokens and create one. You'll see the token exactly once.

    2. Add it to your AI client

    Claude Desktop

    ~/Library/Application Support/Claude/claude_desktop_config.json

    {
      "mcpServers": {
        "lemwatch": {
          "url": "https://trezcnuiaocvjrtiyzho.supabase.co/functions/v1/mcp-server",
          "headers": {
            "Authorization": "Bearer lwm_YOUR_TOKEN_HERE"
          }
        }
      }
    }

    Cursor

    ~/.cursor/mcp.json

    {
      "mcpServers": {
        "lemwatch": {
          "url": "https://trezcnuiaocvjrtiyzho.supabase.co/functions/v1/mcp-server",
          "headers": { "Authorization": "Bearer lwm_YOUR_TOKEN_HERE" }
        }
      }
    }

    For ChatGPT and other clients, point them to https://trezcnuiaocvjrtiyzho.supabase.co/functions/v1/mcp-server with header Authorization: Bearer lwm_…

    3. Try a prompt

    "Use Lemwatch to triage the most recent ongoing incident on lemwatch.com — find the likely root cause and recommend a next action."

    What's exposed

    Tools (11)

    • list_sites
      viewer+
      List monitored sites with current status
    • get_site
      viewer+
      Full site detail with health scores
    • list_incidents
      viewer+
      Incidents filtered by site/status/severity
    • get_incident
      viewer+
      Incident with last 5 checks as evidence
    • list_recent_checks
      viewer+
      Latest uptime probes for a site
    • get_ssl_status
      viewer+
      TLS cert grade, expiry, issuer
    • get_seo_audit
      viewer+
      Latest SEO crawl with issues
    • get_uptime_history
      viewer+
      Daily uptime % and response-time trend
    • compare_sites
      viewer+
      Side-by-side health comparison of two sites
    • list_maintenance_windows
      viewer+
      Scheduled maintenance windows
    • run_pulse_scan
      viewer+
      Run a Production Pulse audit on any URL
    • get_pulse_report
      viewer+
      Fetch a stored Pulse report by id, slug, or domain
    • compare_pulse_scans
      viewer+
      Head-to-head Pulse comparison for two domains
    • acknowledge_incident
      member+
      Mark incident acknowledged (confirm gate)
    • resolve_incident
      member+
      Resolve an incident (confirm gate)
    • trigger_recheck
      member+
      Run all checks now (confirm gate)
    • add_handoff_note
      member+
      Post to team handoff inbox (confirm gate)
    • add_maintenance_window
      admin+
      Suppress alerts during maintenance
    • get_audit_log
      admin+
      Read the MCP request log for this org

    Resources (4)

    Live URIs the agent can subscribe to without burning tokens.

    • lemwatch://sites
    • lemwatch://incidents/active
    • lemwatch://fleet/health
    • lemwatch://pulse/recent
    • lemwatch://sites/{id}/status

    Prompts (5)

    Pre-built investigation playbooks via slash menu.

    • /incident-triage
    • /root-cause
    • /performance-regression
    • /weekly-client-report
    • /pre-launch-check
    • /pulse-audit

    Tier limits

    Free

    Read-only basics

    • list_sites · get_site · list_incidents · get_incident
    • 1 site visible · 30 calls/hour

    Pro · $19/mo

    Full read + writes

    • All tools except maintenance windows
    • 3,600 calls/hour

    Agency · $59/mo

    Everything

    • All tools including maintenance windows
    • 7,200 calls/hour

    Prebuilt agent skills

    Markdown SKILL.md files that teach Claude or Cursor exactly how to use the Lemwatch MCP for triage, pre-launch audits, and weekly client reports.

    For MCP registries & directories

    Lemwatch publishes a standards-compliant discovery manifest at /.well-known/mcp.json so any MCP-aware client or registry can auto-discover capabilities, scopes, and rate limits.

    Manifest URL: https://lemwatch.com/.well-known/mcp.json

    Common questions

    Can I connect my website monitoring to Claude, ChatGPT or Cursor?

    Yes. Lemwatch runs a hosted MCP server over streamable HTTP. Mint a token, paste the server URL and token into your client's MCP configuration, and your assistant can read sites, incidents, SSL, SEO and Pulse audits, and act on them with a confirmation gate.

    What is an MCP server for website monitoring?

    Model Context Protocol is the standard way an AI assistant reads live data from an outside system. A monitoring MCP server exposes uptime, SSL, DNS and performance data as tools the assistant can call, so you ask questions in plain language instead of writing API calls.

    How do I automate website audits for ongoing monitoring?

    Schedule checks in Lemwatch and let alerts run continuously, then use the pulse-audit playbook from your assistant whenever you want a full audit written up. Nothing has to be re-run by hand, and the results stay in one history you can compare against.

    Is there an API as well as the MCP server?

    Yes. The REST monitoring API returns the same uptime, response time, SSL and incident data as JSON and can trigger a check on demand. Full details are on the API documentation page.

    Do I need a paid plan to use it?

    No. The free tier includes MCP access with a reduced tool set and a lower hourly rate limit, so you can connect an assistant before you pay for anything.

    Read the REST API documentation · Browse the agent playbooks

    Create your first token