Code security tools: SAST, SCA, DAST, AI reviews

Code security for software engineers

Topic 9 of 14

Published Nov 26, 2025 · Clip 37:14–47:43 (10:29)

Johannes walks through the categories of code security tools: IDE linting for fast, shallow checks; static application security testing (SAST) using techniques like symbolic execution and taint analysis to model the whole codebase as a graph and trace data flows from user input to security-sensitive sinks; secret detection; infrastructure-as-code scanning; and software composition analysis for known CVEs. He explains the trade-offs of layering tools—choosing developer-oriented tools to avoid noise—and covers dynamic tools: DAST, which shoots malicious payloads at a running application from the outside, and fuzzing for binaries and C/C++ software, noting both are better suited to security teams because of the longer feedback loop. He then gives a nuanced take on AI security reviews: impressive for research agents, but non-deterministic output doesn't work as a consistent quality gate for developers, and using AI to review AI-written code is like students grading their own homework.