Servidor MCP de Atatus

• Direcciónhttps://mcp.atatus.com/mcpno está en el registro oficial: respondió al preguntarle

Ninguna entrada del registro oficial de MCP nombra este servidor. Respondió en atatus.com cuando el Atlas preguntó en la dirección donde suelen montarse los servidores, así que es del propio proveedor. Los clientes y catálogos que listan servidores a partir del registro no lo muestran.

• Estadorespondecomprobado el 2026-10-27
• Herramientas67
• El servidor se llamaAtatus MCP Server
• Versión del servidor1.0.0
• Versión del protocolo2025-06-18

El 2026-10-27 el servidor respondió a initialize y tools/list sin iniciar sesión. Primera respuesta en 94 ms.

• Aún no se ha juzgado si sus herramientas hacen el trabajo del producto.
• Comprobado 2 de los 7 días necesarios para juzgar su fiabilidad.

Lo dicen reglas a partir de las mediciones de arriba. La primera llamada probable la elige un modelo de la lista de herramientas; lee su descripción antes de fiarte.

• Nota INITE MCPtras una semanala nota necesita 7 comprobaciones diarias: 2 hasta ahora
• Respondiómidiendo2 de las 7 comprobaciones diarias necesarias para mostrar el porcentaje
• Mediana de la primera respuesta256 ms
• Espacio en el contexto del agente≈ 32.122 tokenslas definiciones de todas sus herramientas (67), que el agente lleva en cada turno
• Definiciones de herramientasA · 100 de 100
• Resultados tipados0 de 67 herramientasdeclaran outputSchema: el resultado llega tipado y se lee de una vez

• Todas las herramientas pasan las seis comprobaciones.

Un programa comprueba el servidor una vez al día: initialize y tools/list, sin modelo. Los tokens son la longitud de las definiciones, a unos cuatro caracteres por token. La nota promedia seis comprobaciones del nombre, la descripción, el esquema y las anotaciones de cada herramienta; dice lo bien descritas que están, no si funcionan. La nota pondera solo lo medido: si responde (25%), si hace el trabajo del producto (25%), definiciones (20%), contexto (10%), uso (10%) y lo fácil que es conectarse (10%). Lo que no se midió queda fuera, y una instrucción oculta limita la nota a 40. El estado de los servidores MCP de los proveedores: método y números →

• list_projects

  List every project (application) in the Atatus account — returns name, product type, Lang/Framework (subType), and a stable project ID for each. Reads the projects collection, which holds the APPLICATION-type products: APM and RUM/Browser…

• get_apm_metrics

  Project-level Apdex summary for ONE APM project (WEB transactions only): apdex score, successful (error-free) WEB request count, and the satisfied/tolerated/frustrated bucket counts (and their percentages). Targets APM projects (not…

• get_recent_errors

  List recent error groups (deduplicated exceptions) for an APM or browser project. Use this for "what errors are happening right now?", "any new error type today?", "errors since the deploy?". Returns one row per error fingerprint, not per…

• get_error_details

  Get the full story for ONE error group: summary (first/last seen, total events, affected users) plus a few recent events with stack trace + request context + deployment metadata. Use this AFTER get_recent_errors when the user wants "show…

• get_error_events

  List individual occurrences (events) of one error group, each with full stack trace, request context (method/path/status), user info, and agent / release stage. Use this when get_error_details (which returns ~3 events) is not enough — for…

• get_recent_transactions

  List transaction groups (endpoints / background jobs) for an APM project with per-endpoint avg/min/max response time, successful-request count, apdex, total/ok/failed request counts and failure rate. Use this for "which endpoint is…

• get_transaction_spans

  Breakdown of where time is spent INSIDE one transaction — slowest spans/segments classified as DB / External (downstream API) / Internal-App, with response time, call count, and % of total time consumed. The category is derived from each…

• get_apm_database_calls

  APM view of the database calls your instrumented application makes — the APM "Databases" tab, from apm.layers. Returns the database engines your app talks to (PostgreSQL, MySQL, Redis, MongoDB, …) with avg latency, call volume, and share…

• get_recent_deployments

  List the recent deployment markers for a project — version, who deployed, when, environment, repo, revision. Use this for "did we ship anything before the spike?", "what is currently in prod?", "when did v2.4.0 go out?". When to use…

• get_service_health_summary

  One-call "is this service healthy?" — returns a HEALTHY/WATCH/DEGRADED verdict plus golden signals (Apdex, service-wide failed-request count + error rate from HTTP 4xx/5xx, top open error fingerprints, top 5 endpoints by request count…

• diagnose_service

  Composite incident triage for ONE APM service — the "what is wrong with X, and why?" tool to run FIRST when a service is misbehaving. In one call it pulls the golden signals (Apdex + HTTP 4xx/5xx failure rate), the top open error…

• correlate_deploy_with_incident

  Answer "what changed?" — the #1 SRE question during an incident. Given an incident timestamp (or "now" if omitted), list deployments in the lookback window and report the error delta (before vs after each deploy) for the project. Single…

• get_apm_timeseries

  Performance OVER TIME for one APM project: response time, throughput, and HTTP-failure count bucketed across the range, plus a window summary (avg response time, throughput, failure rate/count). This is the data behind the APM dashboard…

• get_error_trends

  Error count OVER TIME for one APM or Browser (RUM) project - error occurrences bucketed across the range, so you can see WHEN errors spiked rather than just which errors exist. Use this for "did errors spike after the deploy?", "is the…

• get_api_analytics

  Break an APM service's real request traffic down by a chosen dimension — country, endpoint (path), customer (companyId), IP, browser, OS, device, domain, host, or request name — with request count, average duration, and failure count/rate…

• update_error_status

  WRITE - change one error group's status: Resolved (mark fixed), Ignored (mute - stops notifications), or Open (reopen). The API-driven equivalent of resolving/ignoring an error in the Atatus UI, so an agent or external system can manage…

• get_kubernetes_overview

  Cluster-wide K8s summary: counts + percentages for Nodes, Pods (running/pending/failed), Containers, Deployments, DaemonSets, StatefulSets, Services, PVs, PVCs, HPAs. Use this as the FIRST K8s tool when the user asks "is the cluster OK?"…

• list_kubernetes_pods

  List Kubernetes pods with status, node, CPU, memory, restart count, and owning workload (Deployment / DaemonSet / Job / CronJob / ReplicaSet). Use this for "which pods are crashing?", "any pod in CrashLoopBackOff?", "what is running in…

• get_kubernetes_pending_pods

  List pods stuck in Pending phase with their (often empty) node assignment and creation time. Use this for "why is pod X not starting?", "any pods stuck Pending?", "is the scheduler healthy?". When to use: scheduling failures…

• list_kubernetes_nodes

  List Kubernetes nodes with capacity (CPU/memory cores, pod allocatable/capacity), age, AND health conditions (Ready, MemoryPressure, DiskPressure, PIDPressure) merged from /nodes + /nodes/status in one call. Use this for "which node is…

• list_kubernetes_deployments

  List Kubernetes Deployments with desired / available / unavailable / up-to-date pod counts and age. Use this for "is rollout X complete?", "any deployment with missing pods?", "show all deployments in namespace Y". When to use…

• get_kubernetes_events

  Get Kubernetes events — the stream of warnings, errors, and state changes across pods, nodes, and other resources. CRITICAL for troubleshooting: why pods failed to schedule, why containers crashed, image-pull failures, OOM kills, and…

• get_kubernetes_unhealthy_workloads

  One-call cross-workload health check: returns the unhealthy Deployments, DaemonSets, and StatefulSets in a single response. Use this for "what is broken across the cluster?", "any workload with missing replicas?". When to use: incident…

• get_pod_details

  Full picture for ONE Kubernetes pod in a single call — state (status, node, cluster, namespace, CPU, memory, restart count, age, owning Deployment/DaemonSet/Job/CronJob) AND recent events for that pod (FailedScheduling, OOMKilled…

• list_pods_on_node

  List pods running on ONE specific Kubernetes node, sorted by CPU or memory. Use this for "which pods are eating node X?", "what is on this node?", "find the noisy neighbor on the hot node". When to use: AFTER list_kubernetes_nodes flags a…

• get_pod_logs

  Fetch logs for a specific Kubernetes pod. Use this immediately after list_kubernetes_pods or get_pod_details surfaces a pod in CrashLoopBackOff, OOMKilled, or Error state — the natural next question is "what did it log before it died."…

• get_infrastructure_overview

  High-level fleet summary: host counts (total / healthy / warning / critical / offline), hot/under-utilized servers, container and process counts, and fleet-average CPU/memory/disk. Use this as the FIRST tool for any "is my infra OK?" /…

• list_infrastructure_hosts

  List monitored hosts (VMs, bare-metal, K8s nodes) with CPU %, memory used/total, disk used/total, and active/inactive status. Returns a fleet summary header plus paginated host rows. Use this for "which host is hot?", "is host X…

• get_infrastructure_checks

  List the up/down/unknown status of agent self-checks, process monitors, and integration (plugin) checks. Use this for "is the agent reporting on host X?", "did the redis/mysql/nginx integration go down?", "which check is failing?". When…

• get_infrastructure_metrics

  Query infrastructure metrics (CPU, memory, disk, network, load, process) for hosts. Use this when list_infrastructure_hosts and get_infrastructure_overview do not have the specific metric you need. Hostname-grouped metrics (CPU, memory…

• list_containers

  List Docker containers across monitored hosts with name, image, host, status, CPU %, memory (RSS), network I/O, and start time. Use this for "which container is eating resources?", "what image is running on host X?", "did container Y…

• get_container_events

  Get the timeline of Docker container lifecycle events — start, stop, restart, OOM kills, exit codes. Use this AFTER list_containers to investigate WHY a container restarted or crashed. Answers questions like "When did payment-api…

• list_host_processes

  List the top processes running on a host with their CPU usage, memory (RSS), memory %, and open file descriptors. Use this when list_infrastructure_hosts shows a host is hot — drill down to find WHICH process is consuming the resources…

• get_host_inventory

  Hardware + OS inventory per host: OS name & version, kernel, distribution, CPU model, CPU cores, RAM, architecture, uptime, boot time. Use this for CVE / patching response ("which hosts run kernel < X?"), compliance audits ("which OS…

• get_logs

  Search RAW application, container, and infrastructure log lines across the entire Atatus account. Works for any source the agent ingests: APM applications, Kubernetes pods, Docker containers, and bare-metal / VM hosts. Filter by hostname…

• analyze_logs

  Summarize log activity for a service or host into the top N recurring patterns with counts. Use this when get_logs would return too many lines to read individually and you want the SHAPE of what is happening: "what error patterns are…

• analyze_kubernetes_event_storm

  Summarize Kubernetes events by reason (FailedScheduling, OOMKilled, ImagePullBackOff, CrashLoopBackOff, BackOff, NodeNotReady, etc.) and return the top N reasons with affected pod / object counts. Use this when get_kubernetes_events would…

• analyze_slow_transactions

  Rank APM transactions for one project by two distinct customer-impact lenses in a single call: (a) HIGHEST TOTAL TIME SPENT = avg response time × throughput (the endpoints actually burning your latency budget); (b) LONG-TAIL OUTLIERS =…

• search_traces

  Search distributed traces across your APM services — find slow or failed requests and see each trace id, total duration, status code, entry service, path, and method. Use for "why is checkout slow?", "show me failed requests in the last…

• get_trace_detail

  Full breakdown of ONE distributed trace — request info, every database call with its query (slowest-first, slowest marked), every remote-service call, error count, and a recomputed time breakdown. The duration is the canonical…

• get_slowest_spans

  The slowest individual spans ACROSS all traces — the most time-consuming database queries, remote-service calls, and operations, aggregated so you see the worst offenders without opening traces one by one. Use for "what is the single…

• get_service_map

  The microservice dependency map for an APM project — every service with its upstream/downstream connections, request rate, error rate, and latency. Use for "map my architecture", "what calls the failing service?", or spotting a cascading…

• get_web_vitals

  Return Core Web Vitals (LCP, FCP, INP, FID, CLS, TTFB) for a BROWSER (RUM) project — average score, the Good / Needs-Improvement / Poor distribution, and an overall status per vital. Core Web Vitals are Google's user-experience +…

• get_page_performance

  Page-load performance for a BROWSER (RUM) project — per page URL: load time, render/network/DOM time, Apdex, error count, and percentiles (P50–P99). Use for "which pages are slow?", "why is /checkout slow for real users?", front-end page…

• get_ajax_performance

  AJAX/XHR (front-end API call) performance for a BROWSER (RUM) project — per endpoint: response time, callback time, throughput, error rate, and data transferred. Use for "which front-end API calls are slow or failing?", client-side…

• get_browser_errors

  Return the distinct JavaScript errors hitting real users in a BROWSER (RUM) project — each grouped error with its message, total occurrences, number of users affected, error type/class, source file:line, and first/last seen. This is the…

• get_browser_error_details

  Drill into ONE browser (RUM) JavaScript error and return its individual real-user occurrences — each with the full JS stack trace (file, line, method), the browser / OS / device / connection, the page URL, the affected user, and geo. This…

• list_databases

  List all monitored database instances (DBM) in the account — each with host, engine type, health, query throughput, average duration, and connections. Use as the STARTING point for database performance work ("which database is unhealthy…

• get_database_health

  Health check for a SINGLE monitored database instance (DBM) — a one-call "is this database healthy right now?" summary: connection status, connection latency, query throughput, average query duration, lock contention, and row-scan…

• get_database_slow_queries

  Rank the slowest database queries by NORMALIZED signature across your monitored databases (DBM) — each row is a query shape with average + total duration, call count, and engine (mysql / postgresql / mongodb / redis / …). This is the…

• get_database_query_details

  Deep detail for ONE database query signature — its EXPLAIN / execution plan, keys examined, average duration, and captured sample executions. Use after get_database_slow_queries to answer "why is this query slow?" and "does it use an…

• list_synthetic_checks

  List all synthetic (uptime) monitoring checks — HTTP, SSL, ICMP, and browser checks, each with status (pass / fail / pause), uptime %, response time, and frequency. Use for "is my site up?", "which endpoints are down?", uptime at a…

• get_synthetic_check_details

  Full detail for ONE synthetic check — configuration, current status, response-time percentiles (avg, p50, p90, p95, p99), and uptime stats. Use to drill into a specific check after list_synthetic_checks ("why is this monitor slow or…

• get_synthetic_check_results

  Individual run results for a synthetic check — each execution with its timing breakdown (DNS, TCP, TLS, TTFB, download), HTTP status, and pass/fail. Use to answer "why is this check failing?" or "where is the latency in the request?"…

• get_availability_report

  Daily uptime report for a synthetic check — availability % and downtime duration for each day in the range (max 100 days). Use for SLA reporting, monthly uptime reviews, and client-facing availability numbers. When NOT to use: per-run…

• list_synthetic_incidents

  Synthetic downtime incidents — when checks detected an outage, with the affected check, start, duration, and status. Use for "what went down and when?", outage history, current active downtime. When NOT to use: current check status →…

• search_topology_entities

  Resolve a human entity name from a prompt into its Flowview `entityKey` (+ type, tier, and owning products). Run this FIRST whenever a question names a thing — "checkout", "payments-db", "the postgres host" — because get_entity_topology…

• get_entity_topology

  The rooted topology map of one entity — run it AFTER search_topology_entities gives you an entityKey. Two modes: • structural (default): the entity's neighbourhood in both directions — its DEPENDENCIES (what it calls / runs on / is…

• get_product_topology

  The account-wide topology of one product — the overview you reach for when a question is about a whole product rather than one named entity. Returns the documented base view: • apm — services and the systems they depend on (databases…

• query_topology_graph

  Run a CUSTOM topology query when the three ready-made tools do not fit — you describe the pattern as a small JSON structure and the server composes valid Cypher from it (so you never write Cypher text and cannot send an invalid query)…

Las primeras 60 de 67, en el orden del servidor.