Skip to content

Detections

Accessing and analyzing CrowdStrike Falcon detections

  • Alerts: READ
  • Alerts: WRITE

Required scopes: Alerts: READ

Count and summarize detections (also called alerts) without retrieving each record.

Use this for “how many” and “top N” questions — alerts per severity, status, tactic, or host, distinct host counts, and alert volume over time — instead of paging through falcon_search_detections. Consult falcon://detections/search/fql-guide before constructing filter expressions. Returns one aggregation per request holding buckets, which key on label with a count; single-value aggregations (cardinality, max, min, avg, sum) report their answer as value instead.

Example prompts:

  • “How many detections do we have by severity?”
  • “What are the top 10 hosts by alert count this week?”
  • “Show me alert volume per day for the last 30 days”
  • “How many distinct hosts have critical alerts?”

Required scopes: Alerts: READ

Retrieve details for detection IDs you already have.

Use when you have specific composite detection ID(s). For discovering detections by criteria (severity, status, hostname, etc.), use falcon_search_detections instead. Returns full detection records; IDs hidden from the Falcon UI are omitted when include_hidden is False.

Example prompts:

  • “Get me the details for this detection”

Required scopes: Alerts: READ

Find detections (also called alerts) by criteria and return their complete details.

Use this to discover detections by severity, status, hostname, time range, or other attributes — this is the tool for general alert and detection queries. Covers alerts across all Falcon products: endpoint (EPP), identity (IDP), XDR, OverWatch, and NG-SIEM. Consult falcon://detections/search/fql-guide before constructing filter expressions. Returns full alert records including process context, device info, tactic/technique details, and threat classification. Responses include pagination.total (the total number of records matching the filter, or null when the API does not report a count) — use it to answer “how many” questions.

Example prompts:

  • “Show me new high severity detections from the last 7 days”
  • “Find all unassigned critical detections”

Required scopes: Alerts: WRITE

Update the status, assignment, visibility, comments, and tags of one or more detections.

Use to change status (new, in_progress, reopened, closed), assign to a user by UUID, email address, or full name, unassign, append a comment, hide/show detections in the UI, or add/remove tags. Resolution is tag-based: applying the conventional tags true_positive, false_positive, or ignored is what populates the console’s Resolution view. At least one update parameter must be provided. Requests covering more than 1000 detection IDs are chunked automatically so none are silently dropped. Returns [] (empty list) on success, or {"result": [], "hint": "..."} when closing without adding a resolution tag in this call; returns an error dict on failure. If a later chunk fails after earlier chunks already applied, the error dict carries a partial_success block listing the already-updated ids so the caller can retry only the remainder rather than re-applying non-idempotent actions.

Example prompts:

  • “Mark detection abc123 as in_progress”
  • “Assign detection abc123 to analyst@example.com
  • “Close these detections and add a comment: resolved via playbook”
  • “Mark detection abc123 as a true positive and close it”
  • “Remove all fc/ prefixed tags from this detection”
  • falcon://detections/search/fql-guide: Contains the guide for the filter param of the falcon_search_detections tool.