Connectors · Observability & analytics

Datadog MCP server for AI agents.

actionlake is an MCP gateway: one remote MCP server that connects Claude and other MCP clients to Datadog and the rest of your systems. It gives your agent 586 Datadog actions (282 reads, 215 writes and 89 that wait for a confirmation) and holds the Datadog credential itself, so the agent never does. It paces Datadog's limits and records every call.

Connect Datadog free See pricing

100,000 free calls a month. No card, no email, no password: press Start a free sandbox on the next screen and keep the key it shows you.

claude mcp add --transport http actionlake https://connect.actionlake.com/mcp

586actionsin one signed file
282readlook, never change
215writecreate and update
89wait for a yesdestructive, confirmed first

Set up

Connect Datadog in three steps.

01
Add actionlake to Claude.
Run the command once in Claude Code; its consent page can create a free sandbox: 100,000 calls a month, no card. In the Claude app, add a custom connector with the URL below and paste your key on the consent page. Creating a new sandbox from the Claude app is not yet tested by us.
https://connect.actionlake.com/mcp
02
Connect Datadog.
Header credentials. No consent screen: your agent can connect it for you, but the key then passes through your chat. To keep it out of the model's context, paste it in the console instead. actionlake stores what you paste write-only, sealed under your tenant's key.
- api_keyDatadog API key
- app_keyDatadog application key
- site_hostWhere your Datadog lives: your site host
```
account { op: "connections.requirements", service: "datadog" }
# free: exactly what to paste, and what is found for you
```

Make the first call. Check before a write.

Ask in plain words; Claude picks the action. Before anything changes, check validates the call and prices it, for free.

do    { action: "datadog.users.list_users_v2" }# list all users. · read · 1 upstream callcheck { action: "datadog.users.disable_user_v1", input: { … } }# free: inputs, permissions, cost and confirmations, nothing sent

Datadog at a glance

Sign-in: Header credentials. No consent screen: your agent can connect it for you, but the key then passes through your chat. To keep it out of the model's context, paste it in the console instead.
You provide: api_keyapp_key
Setup fields: site_host
Limits we pace to: 1,000 calls a minute, per connection
API covered: Datadog API v1+v2, at your Datadog instance's API base URL, which you enter when connecting
Built from: Datadog's documentation, read 18 September 2026OpenAPI spec (GitHub)OpenAPI spec (GitHub)
Service file: Version 1.0.0, signed (Ed25519)
Cached reads: 266 of 282 reads, for up to an hour; a write clears what it affects
Ready-made plans: 10 in the store
Shared entities: 3, so ask can line Datadog up with your other systems
Data: Processed and stored in Sydney, Australia. No US region yet

Capabilities

What an AI agent can do in Datadog.

586 actions in 1 area. Every one carries its effect, so a key can allow the reads and hold the rest.

282readlook, never change
215writecreate and update
89destructivewait for a confirmation

Monitoring
586
282 read · 215 write · 89 destructive
List all users. · Get a list of incidents. · List Slack user bindings. · List all users.

Datadog's own MCP server

Datadog's MCP server, or actionlake?

Datadog runs a remote MCP server that acts with the signed-in user's own Datadog permissions.

Use Datadog's server when one person wants to work in Datadog from their own AI client. Use actionlake when you also want Datadog behind the same endpoint as your other systems: a key per agent with allowed actions and a spending cap, a named person's approval before destructive actions, a sealed record of every call, and one ask across Datadog and another system.

Source: Datadog's documentation, checked 22 October 2026.

Top actions

The Datadog API, shaped for agents.

The ids your agent types, what each does, and the one name it shares with other systems. 280 of Datadog's 296 action names are shared with at least one other connector, so an agent learns them once.

Action	What it does	Effect	Same name in
`datadog.users.list_users_v2`List all users.	List all users.	read	`identity.user.list`474 connectors
`datadog.incidents.list_incidents`Get a list of incidents.	Get a list of incidents.	read	`monitoring.incident.list`19 connectors
`datadog.users.list_users_v1`List all users.	List all users.	read	`identity.user.list`474 connectors
`datadog.hosts.list_hosts`Get all hosts for your organization.	Get all hosts for your organization.	read	`device.endpoint.list`65 connectors
`datadog.users.disable_user_v1`Disable a user.	Disable a user.	destructive	`identity.user.delete`253 connectors
`datadog.incidents.create_incident`Create an incident.	Create an incident.	write	`monitoring.incident.create`13 connectors
`datadog.teams.list_teams`Get all teams.	Get all teams.	read	`identity.group.list`262 connectors
`datadog.teams.create_team`Create a team.	Create a team.	write	`identity.group.create`205 connectors
`datadog.roles.list_roles`List roles.	List roles.	read	`identity.role.list`184 connectors
`datadog.roles.list_role_templates`List role templates.	List role templates.	read	`identity.role.list`184 connectors
`datadog.monitors.list_monitors`Get all monitors.	Get all monitors.	read	`monitoring.monitor.list`32 connectors
`datadog.monitors.create_monitor`Create a monitor.	Create a monitor.	write	`monitoring.monitor.create`27 connectors

Try it

Prompts to paste into Claude.

Each one names the Datadog actions it will call, so you know what it reads and what it changes before you send it.

Readread
Which Datadog incidents are open, and how long has each been open?
callsdatadog.incidents.list_incidents
Planwrite
Onboard a new team with a role, keys and membership in Datadog. Ask me before anything is sent or changed.
callsdatadog.authentication.validatedatadog.roles.clone_roledatadog.users.send_invitationsdatadog.teams.update_team_membershipdatadog.service_accounts.create_service_account_application_key+2 more
Planwrite
Connect a new AWS account to Datadog in Datadog. Ask me before anything is sent or changed.
callsdatadog.aws_integration.list_aws_accountsdatadog.aws_integration.list_aws_namespacesdatadog.aws_integration.create_new_aws_external_iddatadog.aws_integration.create_aws_accountdatadog.aws_integration.create_aws_account_ccm_config+1 more
Writewrite
Create an incident in Datadog. Show me exactly what you will send before you save it.
callsdatadog.incidents.create_incident

The plan store

Ready-made plans for Datadog.

10 plans in the store. Install one and your agent runs it by name; the reply says which connections are still missing. A schedule can run a plan on a timetable. There are no triggers or webhooks, by design.

Build the monthly Datadog cost attribution report
Use it when Finance or platform engineering wants a month-end breakdown of Datadog spend by team or product.
Effect
read
Steps
5
Needs
end month, start hr, start month and tag keys
Install
datadog-monthly-cost-attribution-report
Triage open security signals and escalate the real ones
Use it when A security analyst is working the Datadog signal queue and needs noise archived and real findings escalated.
A person decidesArchive number of data reviewed signals and escalate signal escalate signal id to an incident.
Effect
write
Steps
5
Needs
bulk edit, escalate signal id, incident link and search
Install
datadog-security-signal-triage
Investigate an SLO that has burned its error budget
Use it when An SLO has breached and someone needs the burn history plus the metric and event context behind it.
A person decidesApply an SLO correction to slo id covering the maintenance window, rewriting its recorded history.
Effect
write
Steps
6
Needs
correction, correction id, from ts, metric query, slo id, slo query and to ts
Install
datadog-slo-breach-investigation
Pause synthetic tests for a deploy window
Use it when A release window needs Datadog synthetic tests quiet and then restored automatically.
Effect
write
Steps
6 · loops
Needs
downtime and window
Install
datadog-synthetic-test-maintenance-window
Connect a new AWS account to Datadog
Use it when A new AWS account needs connecting to Datadog with cost collection and a checked metric filter.
A person decidesConnect a new AWS account to Datadog using external id data, starting billable metric collection.
Effect
write
Steps
7
Needs
account and ccm config
Install
datadog-aws-account-integration-setup
Open an incident and wire up its first responders
Use it when A new Datadog incident has been declared and needs impact, responders and triage todos attached.
Effect
write
Steps
7
Needs
impact, incident id, page and responder
Install
datadog-incident-triage-kickoff
Onboard a new team with a role, keys and membership
Use it when A new team is joining Datadog and needs a role, invitations, membership and a service account key.
A person decidesSend Datadog invitations for the new team and grant them the role cloned from template role id as data.
Effect
write
Steps
8
Needs
app key, invitations, membership, role body, service account id, team id, template role id and user id
Install
datadog-onboard-team-to-datadog
Review and revoke stale application keys
Use it when An access review needs old Datadog application keys revoked after sign-off.
A person decidesRevoke every Datadog application key created before cutoff.
Effect
destructive
Steps
5 · loops
Install
datadog-stale-application-key-cleanup
Audit unused dashboards and dashboard lists
Use it when Someone wants to find and clear out Datadog dashboards and lists nobody looks at any more.
A person decidesDelete the empty Datadog dashboard lists found in this audit; number of data cold dashboards are reported only.
Effect
destructive
Steps
6 · loops
Install
datadog-dashboard-usage-audit
Safely decommission monitors for a retired service
Use it when A service is being retired and its Datadog monitors must be removed without breaking composites.
A person decidesDelete number of monitors Datadog monitors matching 'service query'.
Effect
destructive
Steps
6 · loops
Needs
service query
Install
datadog-monitor-decommission-sweep

install a plan

account { op: "plans.install", slug: "datadog-monthly-cost-attribution-report" }
# then ask for it by name; a schedule can run it on a timetable

Governance for Datadog

Destructive actions wait for a yes.

89 Datadog actions are destructive. The first 10:

Disable a user.datadog.users.disable_user_v1
Disable a user.datadog.users.disable_user_v2
Delete an existing incident.datadog.incidents.delete_incident
Remove a team.datadog.teams.delete_team
Delete a monitor.datadog.monitors.delete_monitor
Delete role.datadog.roles.delete_role
Delete an API key.datadog.key_management.delete_api_key_v1
Delete an API key.datadog.key_management.delete_api_key_v2
Delete an On-Call notification channel for a user.datadog.on_call.delete_user_notification_channel
Remove a Slack integration channel.datadog.slack_integration.remove_slack_integration_channel

A destructive action runs only with confirm: true. Under the human policy the agent cannot confirm for itself: the run stops until a named person approves, in the console or from a signed link in an email. The cap on a key is refused before the call is sent, never after.

dodatadog.incidents.delete_incidentconfirm_policy: human

Delete an existing incident.?

intent: the user asked me to delete an existing incident.

run suspended · waiting for a named person · key datadog-agent cannot confirm itself

In the console, or from an emailed signed link.

a key for this connector

account {  op: "keys.create", name: "datadog-agent",  allowed_actions: ["datadog.*"], cap_calls: 2000,  confirm_policy: "human_for_writes",  approval_rules: [    { match: "datadog.users.disable_user_v1", approver: "[email protected]" },    { match: "datadog.users.disable_user_v2", approver: "[email protected]" },    { match: "datadog.incidents.delete_incident", approver: "[email protected]" }  ]}

Illustrative. Scoped to Datadog only, capped at 2,000 upstream calls, every write confirmed, and the destructive ones approved by a named person.

Cost

What Datadog costs on actionlake.

Every Datadog action is one upstream call, the only thing actionlake bills. Searching, check, cost, account and any call Datadog refused with 401, 403, 429 or a 5xx are free. Paid plans start at US$49 a month for 250,000 calls, with overage from US$0.30 down to US$0.18 per 1,000.

1upstream call per actionall 586 Datadog actions
586actions pageeach page of a long list is one call
266reads can be cacheda cache hit costs nothing
100,000calls a month, freepaid plans from US$49 a month for 250,000

See pricing

Action index

Every Datadog action.

586 actions by area: the id your agent calls, what it does, and its effect.

Monitoring586

datadog.incidents.list_incidentsGet a list of incidents.read
datadog.slack_integration.list_slack_user_bindingsList Slack user bindings.read
datadog.users.list_users_v1List all users.read
datadog.users.list_users_v2List all users.read
datadog.audit.list_audit_logsGet a list of Audit Logs events.read
datadog.audit.search_audit_logsSearch Audit Logs events.read
and 580 more in this area

This index shows a sample of each area. All 586 actions, with their inputs and prices, are in the catalogue, which a free sandbox key can search.

Questions

The Datadog MCP server, answered.

Is there a Datadog MCP server for Claude?

Yes. actionlake is an MCP gateway: a remote MCP server that gives Claude, or another MCP client, 586 Datadog actions, through the same eight tools it uses for every other connector. actionlake is independent; it is not made or endorsed by Datadog.

Does Datadog have its own MCP server?

Datadog runs a remote MCP server that acts with the signed-in user's own Datadog permissions. (Datadog's documentation, checked 22 October 2026.) Use Datadog's server when one person wants to work in Datadog from their own AI client. Use actionlake when you also want Datadog behind the same endpoint as your other systems: a key per agent with allowed actions and a spending cap, a named person's approval before destructive actions, a sealed record of every call, and one ask across Datadog and another system.

What do I need to connect Datadog?

Datadog API key and Datadog application key. Plus your site host. The free call account { op: "connections.requirements", service: "datadog" } lists exactly what to paste.

Can an AI agent delete or change things in Datadog?

Only if its key allows it, and only with an explicit confirmation. 89 Datadog actions are destructive, including disable a user., delete an existing incident. and remove a team.. Each runs only with confirm: true. Under the human confirm policy, or an approval rule on the key, the run waits until a named person approves it.

Will an AI agent use up my Datadog API limit?

actionlake paces every call to the limits in the Datadog service file: 1,000 calls a minute per connection. Repeated reads are answered from cache (266 Datadog actions can be), which never reaches Datadog.

What does a Datadog action cost?

One upstream call, the only unit actionlake bills. A list that pages bills one call per page (586 Datadog actions page). Cache hits, searches, dry runs and calls Datadog refused with 401, 403, 429 or a 5xx are free. The Free plan includes 100,000 calls a month with no card. Paid plans start at US$49 a month for 250,000 calls, with overage from US$0.30 down to US$0.18 per 1,000.

Which Datadog API does it cover?

The Datadog API v1+v2 at your Datadog instance's API base URL, which you enter when connecting. It is built from Datadog's documentation as read on 18 September 2026. 586 actions in 1 area, each one in the signed service file (version 1.0.0).

Does actionlake keep my Datadog data, and where?

actionlake processes and stores data in Sydney, Australia; there is no US region yet. Reads may be cached for up to an hour, and a write clears the reads it affects. Each run's request and response is kept, sealed, for your plan's trace window: 1 day on Free, 3 days on Starter, 7 days on Pro, 14 days on Team, 30 days on Business and 30 days on Enterprise. Your Datadog credential is sealed under your tenant's key, never logged and never returned. Your traffic is not used to train models.

Which AI clients can use it?

Claude Code, tested by us end to end, and the Claude app as a custom connector, tested by us with an existing key. Other MCP clients that support remote servers over Streamable HTTP (MCP revisions 2025-06-18 and 2025-11-25) and OAuth dynamic client registration, or a bearer header, should connect the same way; we have not tested them. Or plain REST, under the same action names.

Give your agent Datadog today.

Connect Datadog free See pricing

Start free - 100,000 calls/month, no card. No signup. Hard caps on every key. Cached answers are free.

Datadog MCP server for AI agents.

Connect Datadog in three steps.

Add actionlake to Claude.

Connect Datadog.

Make the first call. Check before a write.

What an AI agent can do in Datadog.

Monitoring

Datadog's MCP server, or actionlake?

The Datadog API, shaped for agents.

Prompts to paste into Claude.

Ready-made plans for Datadog.

Build the monthly Datadog cost attribution report

Triage open security signals and escalate the real ones

Investigate an SLO that has burned its error budget

Pause synthetic tests for a deploy window

Connect a new AWS account to Datadog

Open an incident and wire up its first responders

Onboard a new team with a role, keys and membership

Review and revoke stale application keys

Audit unused dashboards and dashboard lists

Safely decommission monitors for a retired service

Destructive actions wait for a yes.

What Datadog costs on actionlake.

Every Datadog action.

The Datadog MCP server, answered.

More connectors like Datadog

Give your agent Datadog today.