AI Skill Report Card

Defending Against Cyber Attacks

A-83·Sep 27, 2026·Source: Web

Cyber Defense Agent

13 / 15

Given an incident description or a system to defend, produce four sections in this order: Defense Plan → Detection Flow → Response Flow → Recovery Plan. Always ground recommendations in the actual context provided (assets, logs, indicators); do not output generic boilerplate unrelated to the case.

Example trigger: "We see unusual outbound traffic from a web server at 2 AM, possible C2 beaconing."

Recommendation▾
Add an example for a different threat class like ransomware or DDoS to show broader coverage
14 / 15

Progress:

  • Step 1: Clarify scope — what asset/system/network is involved, what evidence exists (logs, alerts, IOCs)
  • Step 2: Classify the threat type (intrusion, malware, phishing, DDoS, insider, data exfiltration, ransomware, etc.)
  • Step 3: Build the Defense Plan — preventive controls, hardening steps, monitoring gaps to close
  • Step 4: Build the Detection Flow — how the threat is/would be identified, in ordered steps (log sources → correlation → alert triggers → triage)
  • Step 5: Build the Response Flow — containment → eradication → communication, in ordered steps with clear owners/actions
  • Step 6: Build the Recovery Plan — restoration steps, validation, post-incident hardening, lessons-learned
  • Step 7: Flag any assumptions made due to missing information
Recommendation▾
Include a severity/triage classification framework (e.g., P1-P4) to guide prioritization decisions

Always structure the final answer as:

  • [Preventive controls, hardening, policy recommendations]
  1. [Step] → 2. [Step] → 3. [Step] ... (Include data sources: SIEM, EDR, firewall logs, IDS/IPS, etc.)
  1. Containment: ...
  2. Eradication: ...
  3. Communication/Escalation: ...
  • Restoration steps
  • Validation/testing before going back to production
  • Post-incident review and hardening actions
18 / 20

Example 1: Input: "Suspicious outbound traffic from a web server, port 443 to unknown IP, at odd hours."

Output:

Recommendation▾
Add guidance on legal/regulatory notification requirements (breach disclosure timelines) as part of Communication step
  • Enforce egress filtering; whitelist known destinations for the web server
  • Deploy EDR agent on the host if not present
  • Segment web server from internal network (DMZ isolation)
  • Enable full packet capture on suspicious segment
  1. SIEM correlation rule triggers on anomalous outbound connection (unusual destination/time)
  2. Cross-reference destination IP against threat intel feeds
  3. Check DNS logs for beaconing pattern (periodic, low-jitter requests)
  4. Confirm via EDR process tree on the host for the initiating process
  5. Escalate to Tier 2 analyst if IOC match found
  1. Containment: Isolate host from network (block egress at firewall, disable NIC if critical)
  2. Eradication: Identify and kill malicious process; remove persistence mechanisms (scheduled tasks, cron, registry run keys)
  3. Communication: Notify incident commander and stakeholders; open incident ticket with severity rating
  • Rebuild host from known-good image if compromise confirmed
  • Rotate credentials/secrets stored/accessible on the host
  • Validate with vulnerability scan + EDR sweep before returning to production
  • Post-incident: review egress rules, add detection rule for this specific C2 pattern, brief team

**Example 2:**
Input: "Employee reports phishing email with a malicious attachment, unsure if opened."

Output:
  • Enforce attachment sandboxing at mail gateway
  • Mandatory security awareness training on phishing
  • Disable macro execution by default org-wide
  1. Retrieve email from mail server logs; extract hash of attachment
  2. Check hash against threat intel / sandbox detonation
  3. Search EDR/SIEM for the hash or related process execution on employee's endpoint
  4. Check for lateral movement indicators if execution confirmed
  1. Containment: Isolate employee's endpoint from network pending analysis
  2. Eradication: If malware confirmed, remove artifacts, block hash/domain org-wide
  3. Communication: Alert affected user, notify IT security team, inform other users if mass phishing campaign detected
  • Reimage endpoint if compromise confirmed
  • Reset employee credentials
  • Verify no further spread via EDR sweep
  • Update email filters with new IOCs; log lessons learned
  • Always tie the Detection Flow to concrete, plausible data sources (SIEM, EDR, firewall, DNS, IDS/IPS) rather than vague statements like "monitor systems."
  • Order Response Flow using standard IR phases: Identification → Containment → Eradication → Recovery → Lessons Learned (Detection/Recovery split out separately here).
  • State assumptions explicitly when input lacks detail (e.g., "assuming no EDR is deployed, recommend...").
  • Prioritize actions by urgency — containment steps first, then deeper investigation.
  • Keep each section actionable — every bullet should be something a responder can execute, not abstract theory.
  • Do not skip straight to "Recovery Plan" without containment — this risks re-infection.
  • Do not give a single generic answer for all attack types; tailor Defense/Detection/Response to the specific threat (ransomware ≠ phishing ≠ DDoS).
  • Do not omit communication/escalation steps in Response Flow — technical fixes without stakeholder notification are incomplete.
  • Do not assume tools (EDR, SIEM) exist without stating the assumption — recommend acquiring them if absent.
  • Avoid vague verbs ("investigate further") without specifying what to check and where.
0
Grade A-AI Skill Framework
Scorecard
Criteria Breakdown
Quick Start
13/15
Workflow
14/15
Examples
18/20
Completeness
18/20
Format
14/15
Conciseness
14/15