# INTRUSIVE.AI > INTRUSIVE.AI (Intrusive AI, Inc.) provides sandboxed attack simulation for agentic AI systems. The suite helps security practitioners connect adversarial input to observable application effects using explicit execution controls and synthetic target environments. Operator Sandbox governs the offensive tester and brokers model and target operations. Attack Harness generates experiments and adapts to permitted feedback. Interceptor Sandbox contains the target application and simulates declared services, files, identities and attacker sinks. Intrusive Threat Intelligence (ITI) is a separate customer-hosted subscription application for threat-intelligence-informed scenario planning and campaign assessment. Operator Sandbox, Attack Harness and Interceptor Sandbox are available under AGPL-3.0-only. Use the release site for version-specific artifacts, compatibility and validation. ITI is subscription software. Operator and Interceptor can be used without ITI. Logical snapshots restore declared state into fresh target processes; they do not guarantee identical subsequent model responses. A model conclusion, an attempted action and an observed effect are different evidence categories. Missing observations remain uncertainty. Configured HTTPS targets do not automatically inherit Interceptor's containment or restore capabilities. External standards explain risks and methodology; first-party artifacts substantiate product behavior. The website's animations and failure example are illustrative, not live telemetry or customer findings. ## Section documents - [Overview](https://intrusive.ai/content/overview.md): entity facts, product/status matrix, glossary, practitioner FAQ and category boundaries. - [Attack surface](https://intrusive.ai/content/attack-surface.md): input-to-effect chain and interpretive OWASP/ATLAS mapping. - [Architecture](https://intrusive.ai/content/system.md): component responsibilities, authority, deployment, provider and integration boundaries. - [Use cases](https://intrusive.ai/content/use-cases.md): eight practitioner workflows and questions covering capability escalation, OWASP evaluation, tool integrations, control validation, regression, remediation, resilience and threat-intelligence-informed red teaming. - [Failure record](https://intrusive.ai/content/failure-record.md): native artifact contracts, complete synthetic investigation, restore prerequisites and evidence limitations. - [Scenarios](https://intrusive.ai/content/scenarios.md): eight scenario families, capabilities, objectives, expected evidence, stopping conditions and coverage gaps. - [Open source](https://intrusive.ai/content/open-source.md): source repositories, AGPL-3.0-only licenses, releases, compatibility and GitHub Issues. - [Trust model](https://intrusive.ai/content/trust-model.md): authority, containment, credential/data flow, administrator-controlled retention and shared responsibility. - [Mission](https://intrusive.ai/content/mission.md): independent authority for sandboxed security testing of AI-based systems, exhaustive testing and radical transparency; supporting vulnerability disclosures, agent incidents and related news distinguish demonstrations from real third-party effects. - [Start](https://intrusive.ai/content/start.md): engagement prerequisites, authorized use, engineering services and contact. ## Complete content - [Combined document](https://intrusive.ai/content/intrusive-ai.md): all section documents combined in page order. ## First-party resources - [Release authority](https://releases.intrusive.ai/): release artifacts, compatibility and validation records. - [Operator Sandbox](https://github.com/intrusiveai/operator_sandbox): campaign controller, host mediation and reports. - [Attack Harness](https://github.com/intrusiveai/attack_harness): adaptive runtime operated through Operator. - [Interceptor Sandbox](https://github.com/intrusiveai/interceptor_sandbox): synthetic target environment, injection controls, snapshots and evidence. ## Optional - [Website](https://intrusive.ai/): visual presentation of the offering. - [Technical briefing](https://intrusive.ai/presentation.html): interactive eight-slide presentation covering the mission, attack surface, suite components, open source, practitioner use cases and available engineering services. - [Website privacy policy](https://intrusive.ai/privacy_policy.txt): website security logs, font requests and email handling; separate from product evidence retention. - [Operator issues](https://github.com/intrusiveai/operator_sandbox/issues): report component security findings and bugs. - [Attack Harness issues](https://github.com/intrusiveai/attack_harness/issues): report component security findings and bugs. - [Interceptor issues](https://github.com/intrusiveai/interceptor_sandbox/issues): report component security findings and bugs.