Intrusive.ai open-sources agent attack simulation sandbox
Understand this story
Intrusive.ai released an open-source agent attack simulation tool on Show HN, using three components—Operator Sandbox, Attack Harness, and Interceptor Sandbox—to put real agent applications into a recoverable synthetic environment for testing. All three components are open-sourced under the AGPL-3.0-only license.
AI overview based on source reports · Updated 2 days ago
Coverage timeline
Follow the coverage to explore different perspectives.
- Hacker News AIShow HN: Intrusive.ai – open-source AI attack simulation
Intrusive.ai is an open-source (AGPL-3.0-only) sandboxed attack simulation platform for agentic systems, splitting an offensive Attack Harness and Threat Intelligence layer from a target agent running in a restorable synthetic world. It records the path from adversarial input to consequential effect across scenarios like indirect prompt injection, capability escalation, credential misuse, and data exfiltration, with declared state that can be restored and branched. The Operator Sandbox, Attack Harness, and Interceptor Sandbox are all open source.
Activity for this story
- Current comparable activity
- 3
- Peak comparable activity
- 10Nov 19 04:00
- Change in the past 24 hours
- -50%
Trends compare the same fully observed participants, which may cover fewer participants than current activity. Move the pointer or tap the chart to inspect hourly activity; use the left and right arrow keys with a keyboard.