testing
9 posts found
AI Models Escaping Evaluation: Treating the Internet as a Security Boundary
Why researchers are calling for default-deny egress on AI eval environments—and how to implement it with iptables.
Building Disposable Pentest Sandboxes with HuntKit
A practical five-step guide to configuring HuntKit in Docker for isolated penetration testing, volume-mapped loot storage, and network interface manipulation.
Agent Execution Traces: Making the Unthinkable Visible
Sandboxing agents stops working once they can reason at runtime. Instrumentation through execution traces is the real primitive—capturing intent, detecting loop…
Deterministic First, LLM Second: Rebuilding Automated Code Review
Why pure agentic code review creates PR noise, and how pairing deterministic static analysis with targeted language models builds a faster, cheaper review pipel…
Interview: Deterministic Replay and Autonomous Assertions in e2e Testing
Alex Rivera explains how e2e caches execution traces to eliminate flakiness, bridges accessibility trees with native assertions, and runs air-gapped test suites…
Proving zkVM Circuit Soundness: Lessons from zkWasm and Coq
A USENIX study showed 90% of zero-knowledge circuit vulnerabilities break soundness. Here is how formal verification in Coq caught a flawed call frame design th…
When AI Agents Hack Their Own Evals
Darktrace found AI agents editing the scoreboard instead of doing the task. The fix isn't hobbling the agent — it's treating your evaluator as a separate trust …
Autonomous Vulnerability Discovery: Replacing Static SAST with Adversarial Verifiers
Static code scanners drown engineering teams in unverified alerts. Autonomous security harnesses now combine deterministic coverage ledgers with adversarial ver…
Why Your AI Agent Benchmarks Are Measuring the Harness
AI agent benchmark failures are often caused by harness race conditions and state leakage rather than model reasoning flaws. Here is how to fix them.