Skip to main content
Version: 0.10

CrowdStrike process events (WIP)

Collects process execution events per endpoint from the CrowdStrike Falcon LogScale (Humio) API and stores them with the scan, one record per endpoint keyed by its Falcon device ID. It is a behavioural record of which executables ran on a host, rather than an inventory source.

Type namecrowdstrike_process_events
CreatesNothing — enriches endpoints that already exist
Runs onA sensor with outbound reach to the Falcon API
Field prefixNone; events are stored as event data, not asset fields
This integration writes no asset fields

It defines no default fields and skips the identification stage entirely — it is an event store. Assets are not created, updated or deleted by it, so its action settings have no effect. Use CrowdStrike for inventory.

What it contributes​

The sensor reads the endpoint list from this deployment's inventory, not from Falcon. For every endpoint that has a cs_device_id (a 32-character Falcon device ID; anything else is skipped), the integration runs a query job against the Falcon investigate_view repository over the Start Window, and collects the results grouped per host and executable image name.

That gives you, per asset, the distinct binaries that executed in the window. The query groups without counting, so it does not say how often each one ran. It is useful for spotting execution of things that have no business being on a host, and for understanding what a machine actually does before you decide how to treat it.

Requirements​

The integration is outbound-only: it is a scheduled query client, so nothing is installed on the CrowdStrike side, nothing runs on the endpoints themselves, and nothing has to reach into the deployment.

Network access​

FromToProtocolPort
The sensor running the integrationThe Falcon API host for your region, e.g. api.us-2.crowdstrike.comHTTPS over TCP443

The token request and every query job go to that same host — the /humio path in the API URL is a route on it, not a separate service — so one egress rule covers the whole integration. Open it for your region's host, and if the sensor reaches the internet through a proxy, point Proxy settings at it rather than opening direct egress.

Access in CrowdStrike​

NeedWhy
An account that can create API clients in the Falcon consoleThe credential is created under Support and resources → API clients and keys, which is admin-gated
Query access to the LogScale investigate_view repositoryThe only Falcon API the run calls besides the token endpoint; the host list comes from the inventory, so no Hosts scope is used. See Permissions in CrowdStrike
A LogScale entitlement that permits query jobsThe events come from investigate_view via query jobs; whether that is available depends on the Falcon subscription

In this deployment​

NeedWhy
A registered, healthy sensorThe integration runs on a sensor, on the platform's schedule — see Sensors
Endpoints already discovered by CrowdStrikeOne query job runs per known cs_device_id; with no CrowdStrike hosts in the inventory there is nothing to query for
Run duration scales with the estate, not the sensor

Because it issues one query job per endpoint, this is the longest-running integration in the catalogue on a large estate. Give it a sensor that stays up rather than the smallest one available, and read the cost warning under Configuration before widening the window.

Configuration​

FieldRequiredNotes
NameYesLabel for this instance
API URLYesThe Falcon API base URL for your region, e.g. https://api.us-2.crowdstrike.com, without a path. The sensor appends /oauth2/token and /humio/api/v1/... itself
Client IDYesFalcon API client ID
Client SecretYesFalcon API client secret. Write-only: once saved it shows as a locked box with Replace, and is sent only to the sensor
Proxy settingsNohttp://host:port
SensorYesSensor that will run the integration
Ignore SSLYesA checkbox. The collector does not currently apply it: certificates are always validated
Leave /humio off the API URL

The console's hint for this field shows an example ending in /humio. The collector adds that path itself, so a URL ending in /humio sends the token request to the wrong address and authentication fails.

Test connection requests a token and then submits a one-minute, single-event query job to investigate_view, cancelling it straight away. It therefore proves query access, not just the credentials.

Unlike other integrations, its advanced settings are specific rather than generic:

SettingDefaultWhat it does
Start Window7dHow far back to look for events
Query Timeout (sec)120Maximum wait for a single query job
Poll Interval (sec)2Delay between polls of a query job
Max Lines / Host1000Cap on events collected per host
Cost scales with hosts × window × lines

One query job runs per endpoint. A wide start window across a large estate is a lot of LogScale work and a long run. Start at the defaults and widen only if the seven-day view is genuinely insufficient.

Permissions in CrowdStrike​

The API client needs read access to the event search / LogScale repository the queries target (investigate_view). No Hosts scope is needed: the endpoints come from the inventory. When access is missing, Test connection names the permission as Investigate (LogScale): Read.

TODO

Confirm the exact Falcon API scope name that grants query-job access to the investigate_view repository for the current Falcon console, and document whether an on-demand LogScale subscription is a prerequisite.

See also​