How to Connect Dynatrace to ChatGPT via MCP

💡 No install? Use cloud.anythingmcp.com directly. Sign in, click Connectors → Dynatrace, paste your credentials, mint an MCP API key and you're done. No Docker, no git clone, no local server.

Dynatrace + ChatGPT

Dynatrace observability through the Environment API v2: Davis problems with root cause and impact, monitored entities and tags, metric queries, log search, events, security problems, SLOs, settings and the audit log. The AI can also comment on problems, send deployment and custom events and tag entities.

Prerequisites

See the full setup instructions baked into the connector (visible in the in-app store when you select the connector). The required environment variables for this connector are:

DYNATRACE_ENVIRONMENT_URL, DYNATRACE_API_TOKEN

Step 1 — Get credentials

Setting up takes about 3 minutes. You need a Dynatrace SaaS or Managed environment.

1. DYNATRACE_ENVIRONMENT_URL is the classic environment address: https://<environment-id>.live.dynatrace.com for SaaS or https://<your-domain>/e/<environment-id> for Managed, without /api and without a trailing slash. The platform address (<environment-id>.apps.dynatrace.com) does not accept this kind of token; use the live address of the same environment.
2. In Dynatrace open Access tokens (Manage, Access tokens, Generate new token) and tick the scopes the tools need: problems.read, entities.read, metrics.read, logs.read, events.read, securityProblems.read, slo.read, settings.read, auditLogs.read; for writes also problems.write (comments, closing), events.ingest and entities.write (tags). A personal access token works the same way.
3. Paste the token (dt0c01.XXXX.YYYY, without Api-Token in front) into DYNATRACE_API_TOKEN and run dynatrace_list_problems to check it. Platform tokens (dt0s16) do not work with this API.

Problems: dynatrace_list_problems with problem_selector: status("OPEN") lists what Davis AI has detected. dynatrace_get_problem with fields: evidenceDetails,impactAnalysis gives the evidence, root cause, impact and affected entities. dynatrace_list_problem_comments reads the comments and dynatrace_add_problem_comment adds one, for example the cause found or the ticket opened.

Topology: dynatrace_list_entity_types and dynatrace_get_entity_type explain the types and their properties. dynatrace_list_entities finds hosts, services, processes, applications, Kubernetes workloads and cloud resources with an entity selector (for example type("SERVICE"),entityName.contains("checkout")), and dynatrace_get_entity shows properties, tags, management zones and relationships. dynatrace_list_tags reads custom tags; dynatrace_add_tags adds tags to every matching entity, and existing tags stay.

Metrics and logs: dynatrace_list_metrics (search with text) and dynatrace_get_metric find the metric key, unit and dimensions; dynatrace_query_metrics reads data points for up to 10 metrics per selector. dynatrace_aggregate_logs counts log records per time slot and attribute to show where errors cluster, then dynatrace_search_logs returns the records, up to 1000 per call.

Events, security, SLOs and audit: dynatrace_list_events, dynatrace_get_event and dynatrace_list_event_types read events; dynatrace_ingest_event records a deployment, configuration change or annotation on entities. dynatrace_list_security_problems with status("OPEN"),riskLevel("CRITICAL","HIGH") and dynatrace_get_security_problem show vulnerabilities with CVEs, risk assessment and affected entities. dynatrace_list_slos with evaluate: true and dynatrace_get_slo give status, error budget and burn rate. dynatrace_list_audit_logs shows who changed what. Settings are read only: dynatrace_list_settings_schemas, dynatrace_get_settings_schema, dynatrace_list_settings_objects and dynatrace_get_settings_object.

Before writing: an ingested alert, error or availability event can open a problem and notify people; comments and tags are visible to everyone in the environment. So the AI shows you the change and waits for your yes.

Switched off at install: dynatrace_close_problem. A closed problem cannot be reopened (Dynatrace opens a new one if the issue comes back), so a workspace admin turns it on when the team needs it.

Selectors: entity selectors look like type("HOST"),tag("env:prod"),mzName("Production"); criteria joined by commas mean AND, and values are case-sensitive. Problem, event, SLO and security problem selectors work the same way. Metric selectors chain transformations: builtin:service.response.time:splitBy("dt.entity.service"):avg:sort(value(avg,descending)):limit(10).

Data formats: times are UTC milliseconds since 1970 (startTime, endTime, timestamp; -1 means still open). Timeframes accept relative values (now-2h, now-7d), milliseconds or ISO 8601; the default is often the last 2 hours, for the audit log the last 2 weeks. Response times are in microseconds, CPU and memory usage in percent, byte counts in bytes; dynatrace_get_metric gives the unit. Entity ids look like HOST-1234ABCD5678EF90; a problem's id differs from its display id (P-2410123).

Paging: lists return nextPageKey; the AI calls again with only next_page_key set and every other filter left out, until it is missing. Log search pages with nextSliceKey, on classic log storage only; on Grail environments the AI narrows the timeframe instead. Short timeframes keep answers small.

Common errors: 401 means a wrong, expired or revoked token, or a platform token; 403 a token without the scope the call needs (the message names it); 404 an unknown id or a wrong environment address; 400 a malformed selector (the message says where); 429 the rate limit, so wait and retry.

The connector does not run Grail or DQL queries: they need a platform token or an OAuth client, not an access token, and the Dynatrace Remote MCP server covers them. It does not change settings and does not remove tags.

Step 2 — Install the adapter

curl -fsSL https://raw.githubusercontent.com/HelpCode-ai/anythingmcp/main/docker-compose.quickstart.yml -o docker-compose.yml
printf 'JWT_SECRET=%s\nENCRYPTION_KEY=%s\n' "$(openssl rand -hex 32)" "$(openssl rand -hex 32)" > .env
docker compose up -d

Step 3 — Add the connector in ChatGPT

Your server URL: in AnythingMCP, open MCP Servers → the server this connector is on, and copy its URL (https://cloud.anythingmcp.com/mcp/…). Use it wherever this guide shows YOUR_SERVER_ID.

ChatGPT adds MCP servers as plugins on its Plugins page, on the web and on paid plans. Developer mode is no longer needed. In a Business workspace only admins and owners can add one; in Enterprise and Edu an admin can grant the Create plugins with MCPs permission.

1. Go to chatgpt.com/plugins, click Add, then Add custom MCP server.
2. Enter a name, e.g. Name: Dynatrace.
3. Under Connection, enter the Server URL https://cloud.anythingmcp.com/mcp/YOUR_SERVER_ID and keep Authentication on OAuth.
4. Select I understand and want to continue, then Create as a plugin. When ChatGPT asks, sign in to AnythingMCP and allow access.
5. Install the new plugin from your personal plugins, then type @ in a chat and pick it.

Available tools

Example prompts

• "Which problems are open in production right now, and what does Davis see as the root cause?"
• "Show me the 5 services with the highest average response time over the last 24 hours."
• "Were there error spikes in the checkout logs in the last hour? Show me a few sample records."
• "Record a deployment of version 1.4.2 on the checkout service. Show me the event before you send it."
• "Which critical or high security problems are still open, and which entities do they affect?"

FAQ

Does the free ChatGPT plan work with custom MCP servers? OpenAI documents custom MCP servers for paid plans only; the Free plan is not on its list.

Next steps

• Connector landing page
• Connect to Claude
• Connect to Cursor
• Connect to Gemini
• Connect to Meta Muse
• Connect to Microsoft Copilot