"An MCP that effectively manages other MCPs" matching MCP connectors:
GET /v1/connectors – MCP directory API referenceMatching Connector Tools:
Read your team's hosted journal from an AI agent: every machine's streams, in hub order.
Monitor metric streams with control charts and alert only on statistically real drift.
Live health leaderboard of x402 paid endpoints, plus an MCP server readiness check
Document operations audit, delivered per API call as an atom service on Conduit Atoms.
Connect AI assistants to PingRelay to monitor websites and APIs, investigate incidents, retrieve uptime and response-time metrics, manage maintenance windows, and access status pages through the Model Context Protocol. Access: Requires a PingRelay paid plan with MCP access and an authorized API key.
Vigilia watches your live app, explains every problem in plain language and writes the fix prompt. Your AI fixes it; Vigilia confirms with real traffic. Vigilia watches web apps that are live with real users, including apps built with Lovable, Bolt, v0, Cursor and Claude Code. It catches errors, failed requests and downtime, explains each problem in plain language, and writes a fix prompt for it. Connect Vigilia to Claude and ask what's broken. Claude lists your open incidents worst first. For each one it reads the plain-language diagnosis, who it affects, the technical evidence (error, page, failing request, stack frames and what the visitor did right before) and Vigilia's fix prompt. In Claude Code or another coding agent, Claude can then fix the cause in your own codebase. Once you've deployed, it marks the incident fixed, and Vigilia watches real traffic to confirm the error stayed away. If it comes back, the incident reopens as a regression. You can also ask how an app is doing (uptime, open incidents, today's traffic), get the exact install snippet for a new project, and check that a live site is running Vigilia. Vigilia never reads or changes your code: the fixing happens in your session, by your AI, in front of you. The only tool that changes anything is marking an incident fixed (the install check just records its result on the app), and you can connect with read-only access. Your data is stored in the EU. Emails, tokens and card numbers are stripped from what visitors' browsers send, and only that scrubbed error context is sent to Anthropic to write the diagnosis. Requires a Vigilia account. The free plan covers one app; reporting backend errors (Supabase, Next.js routes, Workers) is part of Pro.
Hear it from nimo, not your customers. Monitoring that tells you the fix, not just the alert: uptime, Core Web Vitals, cron check-ins, plus Journey Watch for your signup and checkout flows. Free one-off checks, no account needed.
Signed, append-only receipts of an AI agent's decisions and outcomes, with public proof links.
Read a tool-call log from an AI agent run and see which calls failed, looped or were slow.
Fork-replay one recorded span from a failed agent run and get an attribution verdict.
Read-only access to an ET Ducky organization's Windows and Linux agents, alerts, and fleet summary. Four tools list agents, get one agent with its latest health sample, list alerts, and summarize the fleet. Hosted Streamable HTTP endpoint, authenticated with an organization API token sent as a Bearer header.
Signed stage-completeness receipts: proof that a pipeline zero is a real zero, not a failed scan.
Guardr keeps watch on the sites you manage. Check any domain from outside and get an A to F health grade for HTTP headers, TLS, DNS and cookies, with fix steps for configuration findings. Three tools: check_site runs a fresh check of one domain, list_sites lists the sites your account watches and get_site returns one watched site in detail. Read-only: Guardr proposes fixes and never changes a site. Needs a Guardr API key; the first site is free.
Is an AI tool, agent or MCP server alive, and since when? Checks from the outside, with proof.
Live health of 65 agent APIs/MCPs plus a 15-min LLM leaderboard; recommend(task) picks a provider.
Downdetector for AI agents: stop paying for retries that can't succeed. Check once, then report.
Read-only access to an ET Ducky organization's agents, alerts, and fleet summary
Cost per run, tool, model and day for AI agents, from one call or an OpenTelemetry export.
Archive of verbatim errors with root causes and fixes that AI agents search by exact error string.
Ask an agent why a PHP site is slow: every request with its SQL, HTTP calls, errors and N+1.
Free spend report from an agent log, no key. Hosted proxies: caps, audit export, buy calls.
Live health and AI-readable metadata of invokera.com. Demo of an Invokera-hosted MCP server.
Public MCP digital twin with synthetic systems and an agent firewall. No customer data.
An inter-agent graffiti wall for one completely optional trace.
The Google GKE MCP server is a managed Model Context Protocol server that provides AI applications with tools to manage Google Kubernetes Engine (GKE) clusters and Kubernetes resources. It exposes a structured, discoverable interface that allows AI agents to interact with GKE and Kubernetes APIs, enabling them to inspect cluster configurations, retrieve Kubernetes resource YAMLs, monitor operations like cluster upgrades, diagnose issues, and optimize costs—all without needing to parse text output or use complex kubectl commands.
MCP-native AI SRE. Exposes your production OpenTelemetry problems, traces, and logs over the Model Context Protocol, plus an AI remediation loop that opens a reviewed GitHub fix PR and verifies in production (reopening on regression). Tools include list_problems, get_problem, query_traces, detect_anomalies, and request_problem_remediation. Human-in-the-loop by default — the merge button stays yours.
Investigate centralized application logs from AI tools. Fluxtail provides log search, filters and Live Tail through a hosted HTTP MCP server using account-scoped OAuth connections that users can revoke. A Fluxtail account is required. New accounts receive a 14-day no-card trial; paid plans start at $19/month. Website: https://fluxtail.io . Setup and documentation: https://docs.fluxtail.io/ .
Investigate security events and manage the allow/deny lists an analyst acts on.
Cache-bust forensics + TTL-fit verdicts: names the field that busts your prompt cache.
Signed receipts + horizon statements proving whether an event-triggered agent actually woke.
Mezmo MCP is a remote Model Context Protocol (MCP) server that lets AI assistants and IDE chat agents interact with the Mezmo observability platform via the Model Context Protocol. Use it for streamlined observability, log analysis, and root-cause analysis in your favorite tools. Add Mezmo MCP and you can: 🕵️ Run advanced Root-cause analysis over recent logs 📦 List and describe Pipelines 📤 Export and filter Logs with powerful query syntax
What agents recorded when they called an endpoint. Also serves the agent forum. No key.
An MCP server giving access to Grafana dashboards, data and more.
List, inspect and replay the HTTP requests that reached your open tunnl.gg tunnels.
EU AI Act Art-14 runtime oversight: allow / flag / gate-to-human on an agent action, with receipt.
Manage uptime monitors and incidents and search logs, metrics and traces from an AI assistant.
The official MCP server for Lizard (lizard.build). Ship a service, add Managed Postgres, Managed Redis or Managed Object Storage, read logs and metrics, set secrets, scale replicas and attach domains — in plain words. 33 tools over Streamable HTTP, OAuth 2.1; destructive tools require an explicit confirmation argument.
MonoDuty MCP connects compatible assistants to uptime monitoring, job heartbeats and webhook alerts in an on-call workspace. Inspect monitoring coverage, prepare proposals, and create or manage supported resources with separate permissions. Hosted Streamable HTTP endpoint with OAuth. Requires an active verified MonoDuty account, workspace admin access and an API-eligible plan. Docs: https://monoduty.com/docs/mcp
Brazilian hosting operated end to end by the coding agent: create an app, provision PostgreSQL, Redis and object storage, deploy from GitHub Actions, read container logs, measure memory and CPU, and see the cost in BRL before spending it. Irreversible operations require a human approval code that the agent never sees. Prepaid credit, paid by Pix.
API and website monitoring that validates the JSON a response returns, not just its status code. Create endpoint checks, pull incident reports carrying the actual failed payload, and read health scores from your editor.
High-performance array aggregation and metrics clearing engine. Cleans and bucket-groups noisy metric streams via an $O(N)$ single-pass data sweep. Operates natively with the pay-per-call x402 micropayment framework.
Uptime monitoring that doesn't cry wolf: manage monitors, read incidents, record deploys.
Governed access to production AI-agent traces in an existing ClickHouse store.
Open-source agent that observes and fixes your application. Query logs, traces, metrics, incidents.
Cloud hosted Okahu MCP server that helps you manage genAI trace data