Knowledge in SpartanX allows you to define rules, restrictions, and contextual information that guide how the AI agents behave during engagements. This is the equivalent of the terms and conditions you'd give a third-party penetration testing firm — encoded directly into the platform.
What Is Knowledge?
Knowledge entries are free-text inputs that communicate constraints and context to the SpartanX AI agents. They are human-readable and flexible. You can specify:
Network exclusions — e.g., "Never touch 141.1.1.0/16"
Domain restrictions — e.g., "Do not test any third-party API endpoints"
Technical context — e.g., "This application uses 302 redirects" or "All endpoints require an Authorization header"
Functional behavior notes — anything that helps agents understand the environment they're operating in
Restrictions are the most common type. Think of them as the scoping rules you'd give a human pen tester before they begin work.
Engagement-Level Knowledge
When you open an engagement that hasn't been launched yet, navigate to the Knowledge tab. Here, you can add restrictions specific to that engagement. For example, if you're testing only your development environment, you might restrict the scope to a specific domain and exclude your production IP ranges.
As the engagement runs, agents also contribute to the knowledge base. If an agent discovers an unusual application behavior — such as a non-standard redirect pattern or a shared authentication token — it records this as knowledge, making it available to all other agents in the same engagement. This mirrors how human red teams share intelligence during an operation.
Global Knowledge
Global Knowledge is defined at the organizational level and applies to every engagement automatically. You won't need to re-enter standing restrictions each time you launch a new test.
SpartanX includes a default global entry called Safe Checks, which instructs agents to avoid:
Deleting or truncating data
Running destructive commands (e.g., rm -rf)
Any action that could cause irreversible damage to the target environment
You can extend global knowledge with your organization's specific constraints — IP ranges that are always off-limits, third-party services that should never be touched, or shared context that applies to all testing activity.
Two Levels, One System
To summarize:
Global Knowledge — applies to all engagements organization-wide. Set once, enforced always.
Engagement Knowledge — scoped to a specific test. Set per engagement, takes effect only for that run.
Both levels work together. An engagement-level restriction adds to, rather than replaces, global knowledge.
What's Next
The next section covers NodeX Connectors — the virtual images that bring SpartanX's red teaming capabilities to your internal networks.