AI Supply Chain Compromise Assessment Playbook
A Fortune 500 company's SOC received a notification that a third-party software vendor used by 340 of its employees was compromised in a supply chain attack. The vendor's update pushed 6 weeks ago. The SOC needs to assess internal impact before the board briefing in 4 hours.
When to use this playbook
- Use this playbook when the decision looks like the situation above: A Fortune 500 company's SOC received a notification that a third-party software vendor used by 340 of its employees was compromised in a supply chain attack.
- It is a fit when you have source files in hand and need a structured, reviewable analysis — not a generic chat answer about "Supply Chain Compromise Assessment".
- Do not use it as a substitute for licensed, legal, clinical, or authorized official judgment in the domain.
What you'll need
- Vendor software install inventory (all endpoints with version and install date)
- SIEM logs from all endpoints with the vendor software (6 weeks)
- Network flow data for the vendor's update server IP
- EDR process tree for the vendor application on all affected hosts
- Published IOCs from the vendor's breach notification
Attachments: Documents (Documents)
The Prompt
You are a SOC lead assessing internal impact of a supply chain compromise. I am attaching: Work only from the attached source files. If a conclusion is not supported, say so. Produce: 1. Identify all internal endpoints with the compromised software version and whether the malicious update was successfully installed. 2. Cross-reference installed endpoints against the published IOCs: C2 domains, IP addresses, file hashes, and registry keys. 3. Identify any lateral movement from compromised endpoints: new admin account creation, credential harvesting tools, or unusual authentication events. 4. Assess the crown jewel risk: which compromised endpoints have access to sensitive systems (finance, HR, IP, customer data)? 5. Draft the board briefing summary: what we know, what we don't know, what we've contained, and what the 24-hour action plan is. Call out where independent models are likely to disagree, and list follow-up documents a reviewer should request.
What to expect
- Compromised endpoint inventory
- IOC match results by endpoint
- Lateral movement indicators
- Crown jewel exposure assessment
- Board briefing draft with known/unknown/contained framework
Review before you act
- Validate this output against source files before relying on it: Identify all internal endpoints with the compromised software version and whether the malicious update was successfully installed.
- Validate this output against source files before relying on it: Cross-reference installed endpoints against the published IOCs: C2 domains, IP addresses, file hashes, and registry keys.
- Validate this output against source files before relying on it: Identify any lateral movement from compromised endpoints: new admin account creation, credential harvesting tools, or unusual authentication events.
- Validate this output against source files before relying on it: Assess the crown jewel risk: which compromised endpoints have access to sensitive systems (finance, HR, IP, customer data)?.
- Confirm every cited figure, date, counterparty, or requirement against the attached originals — models compress and can drop a qualifier.
- Treat disagreement between models as a review item, especially on classification, materiality, and recommended next action.
- Do not authorize an operational, clinical, legal, credit, or enforcement action solely because the models agree.
Why compare models on this
For Supply Chain Compromise Assessment, running the same attachments across independent models is useful because the hard part is classification and completeness, not fluency. The workflow is already designed to surface compromised endpoint inventory; ioc match results by endpoint; lateral movement indicators; crown jewel exposure assessment. Those are comparison artifacts — they only exist if more than one model runs. Models disagree on blast radius, attribution confidence, and whether a vendor finding is theoretical or exploitable. Those disagreements mark where an analyst should slow down.
See governed multi-model AI on your own prompt
Compare GPT-5, Claude, and Gemini side by side, with human review and a decision record built in.

