Eradication: Removing the Threat for Good
Eradication is where containment's temporary fix becomes a permanent one. In PICERL, this is the phase for removing the threat, not just isolating it. Get it wrong and the same attacker walks back in through the same door.
What Eradication Actually Requires
Containment stops the bleeding. Eradication closes the wound. Skipping straight to "clean and move on" without addressing how the attacker got in is the single most common reason incidents recur.
| Approach | What It Does | What It Leaves Behind |
|---|---|---|
| Symptom Removal | Deletes the malicious file, kills the process, removes the obvious artifact | The vulnerability or access path that let the attacker in the first time, still open |
| Root Cause Removal | Closes the vulnerability, revokes the compromised access, fixes the misconfiguration that was exploited | Nothing exploitable through the same path; attacker has to find a new way in |
Common Eradication Actions
Eradication actions cluster into a few recurring categories. Most real incidents need more than one at once.
| Action Category | Example |
|---|---|
| Malware and Backdoor Removal | Removing a dropped implant and any secondary payload it staged on disk |
| Credential and Secret Rotation | Resetting passwords, API keys, and tokens for every account or service touched during the intrusion |
| Patching the Exploited Vulnerability | Applying the vendor patch or configuration fix for the specific flaw the attacker used to get in |
| Removing Persistence Mechanisms | Deleting malicious scheduled tasks, registry run keys, or rogue services the attacker set up to survive a reboot |
Verifying Eradication
Believing eradication worked and confirming it worked are different things. Verification methods below each catch something self-verification by the same analyst tends to miss.
- Follow-up Scans: catches artifacts the initial cleanup missed, since a single pass rarely finds every dropped file or modified binary.
- Monitoring for Recurrence Indicators: catches a return of the same behavior days or weeks later, when a one-time check would already have moved on.
- Re-checking Persistence Locations: catches a second or backup persistence mechanism the attacker planted as a fallback in case the first one was found.
- Independent Review by a Second Analyst: catches assumptions and blind spots from the analyst who did the eradication work, who is the least likely person to spot their own gap.
When to Rebuild vs Clean
Cleaning saves time. Rebuilding buys certainty. The decision depends on how well the compromise is understood and how much the asset is worth.
| Rebuild | Clean-and-Verify |
|---|---|
| Kernel-level or rootkit compromise | Well-understood, commodity malware |
| Unknown persistence depth | Clear, complete removal path |
| High-value asset where certainty matters most | Low-value asset that's easy to reimage later if something was missed |
Coordinating Eradication Timing Across Systems
When multiple systems are compromised, eradication needs to happen at the same time everywhere, not one system at a time.
Key Takeaways
- Eradication has to address root cause, not just remove the visible symptom, or the same intrusion path stays open.
- Common eradication actions cover malware and backdoor removal, credential and secret rotation, patching the exploited vulnerability, and removing persistence mechanisms.
- Verification should come from more than one method: follow-up scans, recurrence monitoring, re-checking persistence locations, and independent review all catch different gaps.
- Rebuild when the compromise depth is unknown, involves kernel-level access, or the asset is high-value; clean-and-verify when the malware and removal path are well understood.
- Simultaneous eradication across all affected systems prevents the attacker from shifting to systems not yet cleaned.
- Eradicating one system at a time can tip off the attacker and turn cleanup into an ongoing chase.
Knowledge Check
Click an answer to reveal the explanation.
An analyst deletes a malicious executable from an infected host and closes the incident. A week later, the same host is compromised again through the same unpatched service. What went wrong?
A forensic review finds a rootkit with kernel-level hooks on a high-value production server, and the full extent of its persistence is unclear. What's the appropriate eradication approach?
A team eradicates malware from one compromised server, then moves on to the next server the following day. Why is this sequencing risky?