AgentViva — Claude Code MCP server: monitoring and setup
AgentViva — an MCP server for Claude Code checks whether AI tools and services respond. Agents can create monitors, run checks, and read status changes.
MIT updated 2026-11-14
claude mcp add --transport http agentviva https://agentviva.dev/mcpWhat AgentViva does
AgentViva tracks websites, JSON endpoints, and MCP servers, keeping evidence from each check. For a page, it follows redirects; for JSON, it tests rules against the response body; for an MCP target, it can verify that selected tools are listed. It can also send a prompt to an AI service and assess the reply.
Agents can manage monitors, inspect recent changes, and sync tool catalogs through MCP. Sign-in uses a browser approval flow; clients that cannot open a browser can send an API key. One failed probe does not change a status: the same verdict must recur ten minutes later. AgentViva runs as a hosted service, so a local server is not part of setup.
What it helps with
- check whether a site still responds after redirects
- flag a JSON endpoint when a field breaks a rule
- inspect which tools an MCP server exposes
- read recent changes to monitored services
Example prompts for the agent
- Check whether this site responds after redirects and show the result.
- Monitor this JSON endpoint and require
statusto equalok. - List recent changes and monitors that need a decision.
How to install AgentViva
Platform CLI: remote server
claude mcp add --transport http agentviva https://agentviva.dev/mcpPlatform CLI: remote server
codex mcp add agentviva --url https://agentviva.dev/mcpConfig file
{
"mcpServers": {
"agentviva": {
"url": "https://agentviva.dev/mcp"
}
}
}Config file
{
"servers": {
"agentviva": {
"type": "http",
"url": "https://agentviva.dev/mcp"
}
}
}Platform CLI: remote server
gemini mcp add -t http agentviva https://agentviva.dev/mcpStep by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about AgentViva
How do I install AgentViva in Claude Code?
Run claude mcp add --transport http agentviva https://agentviva.dev/mcp. Then open /mcp, select agentviva, and authenticate.
What does AgentViva do?
It checks websites, JSON APIs, and MCP servers, then records their state with evidence from the latest probe. An agent can manage monitors and read changes through MCP.
What does AgentViva need to work?
The service is hosted remotely. Sign in by approving access in a browser; a client without browser access can send an API key.