Firewalls, VPN gateways and routers rarely run EDR, so when one is suspected of compromise the evidence lives in volatile memory, device logs and vendor diagnostics that a reboot or upgrade can destroy. This playbook is a ready-to-adapt procedure for investigating a suspected compromise of an edge appliance: when to start, who does what, how to capture evidence before touching the device, how to contain without tipping off the attacker, and how to rebuild and rotate every secret the device held. It draws on CISA's emergency directive procedures for Cisco ASA and Ivanti Connect Secure and on Fortinet's published persistence analysis. It is for incident responders, network teams and security leads, and it complements the VPN Security Guide, which covers prevention and hardening.
How to use this template
- Adapt before you need it: Replace every [Placeholder] with your names, contacts and tools, and add a vendor annex for each appliance family you run (collection commands, integrity tools, TAC or support contacts).
- Follow the order: Sections 3 to 6 are sequenced so that evidence is captured before containment actions change device state. Departing from the order needs the incident lead's approval and a note in the decision log.
- Vendor procedures win: When the vendor or a government agency publishes collection instructions for a specific campaign, follow them exactly. CISA's ED 25-03 core dump instructions, for example, warn that steps must be followed in order and that tab autocompletion may crash the device.
- Exercise it: Run the playbook on a lab appliance at least once a year so the team knows how long a support bundle or memory capture takes and where it lands.
1. Scope and triggers
1.1 This playbook applies to [Organisation]'s internet-facing and boundary appliances: firewalls, SSL and IPsec VPN gateways, routers, load balancers, secure web gateways, email gateways, file transfer appliances and out-of-band controllers such as iLO and iDRAC.
1.2 Activate it when any of the following occurs:
- A vendor, CISA, NCSC or another national authority reports active exploitation of a product version [Organisation] runs, or a vulnerability is added to the CISA Known Exploited Vulnerabilities catalogue.
- A vendor integrity tool, AV/IPS signature or health check reports an unexpected file, process or failed verification.
- Configuration drift: a new local administrator, a changed AAA server address, new SSH keys, altered ACLs or an unexpected logging target.
- Anomalous behaviour: unexplained reboots or crashes, outbound connections from the device itself, logins from unusual addresses, or log gaps.
- Downstream signals: suspicious activity from the VPN address pool or from accounts the appliance uses to bind to the directory.
1.3 Severity: treat a confirmed indicator on any internet-facing appliance as [Severity 1]. Treat an exploited-vulnerability notice without indicators as [Severity 2] until evidence collection is complete.
2. Roles and contacts
| Role | Responsibility | Named person and deputy |
|---|---|---|
| Incident lead | Owns decisions, sequence and the decision log | [Name], [Deputy] |
| Network engineer | Runs collection commands under direction, knows topology and failover | [Name], [Deputy] |
| Forensic analyst | Defines what to collect, hashes and stores evidence, analyses artefacts | [Name or IR retainer] |
| Identity owner | Rotates directory, service account and cloud credentials the device used | [Name] |
| Vendor support | Integrity tools, memory analysis, TAC case | [Vendor, contract ID, phone] |
| Communications and legal | Regulator, customer and partner notification decisions | [Name] |
| Government contact | Reporting and analysis support (for example CISA, NCSC) | [Contact route] |
2.1 Use an out-of-band channel if the appliance carries the organisation's remote access or email, because an attacker on the device may see normal traffic. See the Out-of-Band Communications for Incident Response Guide and the Incident Commander Role Guide.
3. Detection and triage
3.1 Do not reboot, upgrade or factory reset yet. The VPN Security Guide records vendor warnings that upgrades and reboots can make earlier logs and memory artefacts inaccessible.
3.2 Record device identity: model, serial number, software and boot loader versions, uptime and last crash information. CISA's ED 25-03 instructions record model, serial and version first, then collect show tech-support detail off-device.
3.3 Capture evidence in this order, saving every file off-device and recording a SHA-256 or SHA-512 hash on the device and again after transfer, as the ED 25-03 instructions require for the memory segment dump:
| Order | Artefact | Why it matters | Example |
|---|---|---|---|
| 1 | Running configuration and diff against the last approved backup | Shows added accounts, keys, ACLs, AAA and logging changes | Configuration export, compared with the repository copy |
| 2 | Active sessions, processes, connections and routing | Volatile, lost on reload | Session tables, VPN user list, connection table |
| 3 | Vendor diagnostic bundle | Collects logs and system state in a format vendor support can read | Cisco show tech-support detail; Palo Alto Networks tech support file |
| 4 | Integrity or heap checks the vendor specifies | Some implants show as anomalies in health counters | ED 25-03: run show checkheaps twice at least five minutes apart and check the run counter increases |
| 5 | Core dump or memory image, if the vendor procedure includes one | Only source of evidence for memory-resident implants | ED 25-03 core dump, which forces an immediate reload: schedule with the network team |
| 6 | Off-device logs: syslog, SIEM, NetFlow, AAA accounting, firewall logs of traffic to and from the device | Attackers clear on-box logs; external copies survive | Export the incident window plus [30] days before first indicator |
3.4 Run the vendor's integrity tool after evidence capture, because some tools reboot the device. Treat a clean result as one data point only. CISA advisory AA24-060B reported that Ivanti's Integrity Checker Tool did not detect compromise in some cases, as summarised in the Ivanti Connect Secure exploitation page.
3.5 Look for the persistence patterns seen in real campaigns: new local accounts, added SSH keys, repointed AAA servers and changed ACLs (reported in the Salt Typhoon telecom intrusions); web shells on management or portal components; files planted in user-writable paths, such as the symbolic link in the FortiGate SSL-VPN language file folder that Fortinet described in April 2025 and that survived patching; and modified boot or ROM images, as in the ArcaneDoor-linked Cisco ASA campaign behind ED 25-03.
3.6 Decide: compromise confirmed, suspected (anomalies without a definite indicator) or not found. Record the reasoning and evidence in the decision log. Suspected cases proceed as confirmed for containment and credential rotation.
4. Containment
4.1 If compromise is confirmed, isolate the device from the network but do not power it off. CISA's ED 25-03 gives exactly this instruction for compromised ASA devices.
4.2 Where the appliance provides essential remote access, fail over to a standby or alternative path built from known-good media before isolation, and record the service impact accepted by [Business owner].
4.3 Terminate all VPN and administrative sessions, and block the attacker infrastructure identified in triage at an upstream control rather than on the suspect device. Follow vendor or agency advice where it says to consult before blocking, as the ED 25-03 instructions do.
4.4 Hunt beyond the device. The ED 24-01 supplemental direction for Ivanti told agencies to continue hunting on systems connected to the affected devices, monitor exposed authentication and identity services, and audit privileged accounts. Prioritise the VPN address pool, the directory bind account and any jump hosts reachable from the device. See the Incident Scoping and Containment Strategy Guide.
5. Eradication
5.1 Do not trust in-place patching to remove an attacker. Fortinet's April 2025 analysis showed persistence that survived upgrades to fixed versions, and ED 25-03 describes ROM modification that survives reboots and upgrades.
5.2 Export configuration, factory reset or reimage from vendor media whose hash you have verified, upgrade to a fixed and supported version, then reimport configuration. This is the sequence in the ED 24-01 supplemental direction. Review the exported configuration line by line before reimport, and remove anything not in the approved baseline.
5.3 Replace hardware that is end of support or where the vendor or analysis indicates firmware or ROM persistence.
5.4 Rotate every secret the device held or saw. The ED 24-01 supplemental direction lists certificates, keys and passwords including the admin enable password, stored API keys and passwords for local users and service accounts used to configure authentication servers, and told agencies to assume connected domain accounts were compromised.
| Secret | Action | Owner |
|---|---|---|
| Local admin and enable passwords | Reset with strong hashing; vault the emergency account | [Network] |
| Directory bind and service accounts | Reset; for on-premises accounts tied to the device, ED 24-01 says reset twice and revoke Kerberos tickets | [Identity] |
| Cloud and hybrid identities | Revoke tokens; disable cloud-joined or registered devices to revoke device tokens, as ED 24-01 directs | [Identity] |
| TLS and VPN certificates, private keys | Revoke and reissue with new key pairs | [PKI] |
| AAA shared keys, SNMP credentials, routing and NTP keys | Rotate on every peer, not only the device | [Network] |
| API keys and automation tokens | Revoke and reissue with narrower scope | [Platform] |
| User credentials that passed through the VPN | Force reset where credential harvesting is suspected; review MFA registrations | [Identity] |
5.5 If the investigation shows the attacker used device credentials to reach the directory, escalate to the Active Directory Domain Compromise Response Playbook or the Identity Provider Tenant Compromise Playbook.
6. Recovery
6.1 Reconnect only after: the device runs a verified fixed image, configuration matches the approved baseline, all secrets in 5.4 are rotated, off-device logging is confirmed working, and management access is restricted to the management network.
6.2 Watch the first boot where the vendor gives guidance. CISA's ED 25-03 instructions describe boot loader and ROMMON verification messages, and treat "verification failed" messages as possible persistence.
6.3 Apply heightened monitoring for [30] days: alerts on any configuration change, new accounts or keys, outbound connections from the device, and authentication from the VPN pool to sensitive systems.
6.4 Re-run the vendor integrity tool and compare configuration against the baseline daily during the monitoring period.
7. Communication and notification
7.1 Notify internal stakeholders through the out-of-band channel, keeping technical indicators to the response team until containment.
7.2 Assess regulatory and contractual notification with [Legal]: personal data exposure (for example under UK GDPR or EU GDPR), sector rules, and customer contracts. Record the decision and time in the decision log.
7.3 Report to the vendor's PSIRT and, where appropriate, to the national authority. CISA's ED 25-03 notes that entities outside the federal executive branch may also follow its core dump instructions, and its supplemental direction tells non-federal US organisations with indicators to contact CISA.
7.4 Share indicators with peers through [ISAC or sharing community] once cleared by [Incident lead].
8. Post-incident review
8.1 Hold a review within [10] working days. Answer: how was the device first accessed, how long was the attacker present, what secrets were exposed, and what downstream access followed.
8.2 Measure: time from trigger to evidence capture, time to isolation, time to full secret rotation, percentage of secrets rotated within [72] hours, and whether off-device logs covered the full dwell time.
8.3 Feed fixes into the hardening baseline in the Network Device Management Plane Hardening Guide and the appliance replacement plan. Record findings in the Tabletop Exercise After Action Report Template format.
Implementation guidance
- Prepare collection in advance: For each appliance family, document the diagnostic bundle, memory capture and integrity tool commands, where files are written, how long they take and whether they reload the device. Store copies of vendor collection guides offline.
- Keep evidence off the box every day: Off-device syslog, AAA accounting and flow records over encrypted transport, retained for at least [12] months, are what let investigators scope dwell time.
- Keep a secrets register per device: List every credential, certificate and key each appliance holds, its owner and how to rotate it. Without it, step 5.4 takes weeks.
- Evidence handling: Hash every artefact, record who collected it and when, and store it in a restricted location. See the Evidence Handling and Chain of Custody Guide and the Chain of Custody Form Template.
- Common mistakes to avoid: Rebooting or upgrading before capture; trusting a single clean integrity check; rotating the device password but not the directory bind account or certificates; patching without hunting downstream; and reconnecting before off-device logging is restored.
- Framework fit: The playbook implements NIST SP 800-53 IR-4 Incident Handling, IR-5 Incident Monitoring, IR-6 Incident Reporting and SI-7 integrity verification, and NIST CSF 2.0 outcomes RS.AN-03 (analysis to establish what happened and root cause), RS.CO-02 and RC.RP-01. Relevant ATT&CK techniques include T1190 Exploit Public-Facing Application, T1133 External Remote Services, T1505.003 Web Shell, T1601 Modify System Image and T1556.004 Network Device Authentication.
Related NHI Mgmt Group resources: Edge Device Security Guide · VPN Security Guide · Network Forensics Guide · Memory Forensics Guide · Incident Response Plan Template