Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Deploys and configures CrowdStrike Falcon EDR agents across enterprise endpoints to enable real-time threat detection, behavioral analysis, and automated response. Use when onboarding endpoints to EDR coverage, configuring detection policies, or integrating Falcon telemetry with SIEM platforms. Activates for requests involving CrowdStrike deployment, Falcon sensor installation, EDR policy configuration, or endpoint detection and response.
.claude/skills/deploying-edr-agent-with-crowdstrike/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 91% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 54% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 108% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 75% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 98% | 0% |
Use this skill when:
Do not use this skill for deploying other EDR solutions (Carbon Black, SentinelOne) or for Falcon cloud workload protection (use cloud-specific deployment guides).
1. Log into Falcon Console: https://falcon.crowdstrike.com
2. Navigate: Host setup and management → Sensor downloads
3. Download the appropriate installer:
- Windows: WindowsSensor_<version>.exe
- macOS: FalconSensorMacOS_<version>.pkg
- Linux: falcon-sensor_<version>_amd64.deb / .rpm
4. Copy the Customer ID (CID) from the Sensor downloads page
- CID format: <32-char-hex>-<2-char-checksum>Silent installation via command line:
cmdWindowsSensor_7.18.17106.exe /install /quiet /norestart CID=<YOUR_CID>
SCCM deployment:
1. Create an Application in SCCM
2. Deployment type: Script Installer
3. Install command: WindowsSensor_7.18.17106.exe /install /quiet /norestart CID=<CID>
4. Detection method: Registry key exists
- HKLM\SYSTEM\CrowdStrike\{9b03c1d9-3138-44ed-9fae-d9f4c034b88d}\{16e0423f-7058-48c9-a204-725362b67639}\Default
5. Deploy to target collection
6. Deployment purpose: Required (for mandatory installation)Microsoft Intune deployment:
1. Navigate: Devices → Windows → Configuration profiles
2. Create Win32 app deployment
3. Upload .intunewin package (wrapped sensor installer)
4. Install command: WindowsSensor_7.18.17106.exe /install /quiet /norestart CID=<CID>
5. Detection rule: File exists C:\Windows\System32\drivers\CrowdStrike\csagent.sys
6. Assign to device groupGPO deployment:
powershell# Create startup script that checks for existing installation $sensorPath = "C:\Windows\System32\drivers\CrowdStrike\csagent.sys" if (-not (Test-Path $sensorPath)) { Start-Process -FilePath "\\fileserver\CrowdStrike\WindowsSensor.exe" ` -ArgumentList "/install /quiet /norestart CID=<CID>" -Wait }
bash# Debian/Ubuntu sudo dpkg -i falcon-sensor_7.18.0-17106_amd64.deb sudo /opt/CrowdStrike/falconctl -s -f --cid=<YOUR_CID> sudo systemctl start falcon-sensor sudo systemctl enable falcon-sensor # RHEL/CentOS sudo yum install falcon-sensor-7.18.0-17106.el8.x86_64.rpm sudo /opt/CrowdStrike/falconctl -s -f --cid=<YOUR_CID> sudo systemctl start falcon-sensor sudo systemctl enable falcon-sensor # Verify sensor is running and connected sudo /opt/CrowdStrike/falconctl -g --rfm-state # Expected output: rfm-state=false (sensor is communicating with cloud)
bash# Install sensor package sudo installer -pkg FalconSensorMacOS_7.18.pkg -target / # Set CID sudo /Applications/Falcon.app/Contents/Resources/falconctl license <YOUR_CID> # Grant Full Disk Access and System Extension via MDM profile # Required for macOS Ventura+ (manual approval or MDM PPPC profile) # MDM payload: com.crowdstrike.falcon.Agent → SystemExtension + Full Disk Access # Verify sensor status sudo /Applications/Falcon.app/Contents/Resources/falconctl stats
In Falcon Console, navigate to Configuration → Prevention Policies:
Recommended prevention policy settings:
Machine Learning:
- Cloud ML: Aggressive (extra protection, may increase false positives)
- Sensor ML: Moderate
- Adware & PUP: Moderate
Behavioral Protection:
- On Write: Enabled (detect malware on file creation)
- On Sensor ML: Enabled
- Interpreter-Only: Enabled (detect script-based attacks)
Exploit Mitigation:
- Exploit behavior protection: Enabled
- Memory scanning: Enabled (detects in-memory attacks)
- Code injection: Enabled
Ransomware:
- Ransomware protection: Enabled
- Shadow copy protection: Enabled
- MBR protection: EnabledCreate separate policies for:
Real-Time Response:
- Enable RTR for all sensor groups
- Configure RTR admin vs. RTR responder roles
- Enable script execution (for IR teams)
- Enable file extraction (for forensics)
Network Containment:
- Pre-authorize containment for specific host groups
- Configure containment exclusions (allow management traffic)
Automated Response:
- Enable automated remediation for high-confidence detections
- Configure kill process action for ransomware detections
- Enable quarantine for malware file detectionspowershell# Windows: Check Falcon sensor status sc query csagent # Expected: RUNNING # Check sensor version reg query "HKLM\SYSTEM\CrowdStrike\{9b03c1d9-3138-44ed-9fae-d9f4c034b88d}\{16e0423f-7058-48c9-a204-725362b67639}\Default" /v AgentVersion # Verify cloud connectivity # In Falcon Console: Host Management → Hosts → search for hostname # Status should show "Online" with last seen timestamp < 5 minutes
Test detection capability:
powershell# CrowdStrike provides test detection samples # Download CsTestDetect.exe from Falcon Console → Host setup # Run on endpoint to generate a test detection .\CsTestDetect.exe # Verify detection appears in Falcon Console within 60 seconds
# Falcon SIEM Connector (Streaming API)
# Configure in Falcon Console: Support → API Clients and Keys
# Create API client with scope: Event Streams → Read
# Use falcon-siem-connector or Falcon Data Replicator (FDR)
# Splunk integration:
# Install CrowdStrike Falcon Event Streams Technical Add-on from Splunkbase
# Configure: Settings → Data inputs → CrowdStrike Falcon Event Streams
# Enter API Client ID and Secret
# Index: crowdstrike_events
# Elastic integration:
# Use Elastic Agent with CrowdStrike module
# Configure: Fleet → Agent policies → Add integration → CrowdStrike| Term | Definition | |------|-----------| | Falcon Sensor | Lightweight kernel-mode agent (25-30 MB) that collects endpoint telemetry and enforces prevention policies | | CID (Customer ID) | Unique identifier that associates the sensor with your CrowdStrike Falcon tenant | | RFM (Reduced Functionality Mode) | State where sensor operates with limited capability due to cloud connectivity loss | | Sensor Grouping Tags | Labels applied during installation to auto-assign hosts to groups and policies | | RTR (Real-Time Response) | Remote shell capability for incident responders to interact with endpoints through Falcon | | IOA (Indicators of Attack) | Behavioral detections based on adversary techniques rather than static signatures |
/install /quiet CID=<CID> APP_PROXYNAME=proxy.corp.com APP_PROXYPORT=8080.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 13,932 | 9,714 | -30% | 1 | 1 | 0% | 2,191 | 4,179 | +91% | 0 | 0 | — |
case-02 | fail→pass | 17,289 | 13,021 | -25% | 1 | 1 | 0% | 2,952 | 4,555 | +54% | 0 | 0 | — |
case-03 | fail→fail | 8,806 | 5,673 | -36% | 1 | 1 | 0% | 1,462 | 3,324 | +127% | 0 | 0 | — |
case-04 | fail→fail | 15,264 | 9,557 | -37% | 1 | 1 | 0% | 2,680 | 3,994 | +49% | 0 | 0 | — |
case-05 | fail→fail | 15,326 | 14,147 | -8% | 1 | 1 | 0% | 2,761 | 4,986 | +81% | 0 | 0 | — |
case-06 | fail→pass | 12,886 | 17,230 | +34% | 1 | 1 | 0% | 2,082 | 4,331 | +108% | 0 | 0 | — |
case-07 | fail→pass | 11,075 | 6,920 | -38% | 1 | 1 | 0% | 2,072 | 3,630 | +75% | 0 | 0 | — |
case-08 | pass→pass | 9,353 | 5,579 | -40% | 1 | 1 | 0% | 1,646 | 3,303 | +101% | 0 | 0 | — |
case-09 | fail→fail | 10,590 | 8,768 | -17% | 1 | 1 | 0% | 2,026 | 3,917 | +93% | 0 | 0 | — |
case-10 | fail→fail | 5,248 | 2,350 | -55% | 1 | 1 | 0% | 1,034 | 2,680 | +159% | 0 | 0 | — |
case-11 | fail→fail | 12,359 | 5,258 | -57% | 1 | 1 | 0% | 1,892 | 3,256 | +72% | 0 | 0 | — |
case-12 | fail→fail | 8,291 | 5,220 | -37% | 1 | 1 | 0% | 1,384 | 2,961 | +114% | 0 | 0 | — |
case-13 | fail→pass | 8,300 | 3,357 | -60% | 1 | 1 | 0% | 1,492 | 2,953 | +98% | 0 | 0 | — |
case-14 | fail→fail | 8,565 | 5,532 | -35% | 1 | 1 | 0% | 1,513 | 3,299 | +118% | 0 | 0 | — |
case-15 | fail→pass | 16,043 | 3,728 | -77% | 1 | 1 | 0% | 2,719 | 2,887 | +6% | 0 | 0 | — |
case-16 | fail→fail | 13,544 | 13,569 | +0% | 1 | 1 | 0% | 2,296 | 4,540 | +98% | 0 | 0 | — |
case-17 | fail→pass | 12,816 | 8,274 | -35% | 1 | 1 | 0% | 2,076 | 3,635 | +75% | 0 | 0 | — |
case-18 | fail→fail | 5,088 | 3,119 | -39% | 1 | 1 | 0% | 821 | 2,763 | +237% | 0 | 0 | — |
case-19 | fail→fail | 7,799 | 7,620 | -2% | 1 | 1 | 0% | 1,679 | 3,495 | +108% | 0 | 0 | — |
case-20 | fail→fail | 5,655 | 3,030 | -46% | 1 | 1 | 0% | 1,024 | 2,870 | +180% | 0 | 0 | — |
case-21 | fail→fail | 10,692 | 5,575 | -48% | 1 | 1 | 0% | 1,744 | 3,167 | +82% | 0 | 0 | — |
case-22 | fail→fail | 9,309 | 7,835 | -16% | 1 | 1 | 0% | 1,553 | 3,651 | +135% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +32 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.