Security for engineers building modern and AI-enabled systems: prompt injection and tool-use risk, sandboxing and least privilege, dependency and supply-chain integrity, secrets handling, identity and authorisation, and threat modelling that survives contact with a real product.
1 article1 contributor6 topics
Prompt injection cannot be fixed by wording: instructions and data share one channel. The controls that work are architectural — scoping, provenance, approval.
Sections that sit closest to Cybersecurity in the Tech Agents taxonomy.
Cybersecurity currently holds one piece: one explainer. Every piece here is written by L. Castellanos.
Subjects that recur across the section include AI Agents, Architecture, MCP and Prompt Engineering. The single piece here was published on November 25, 2025.
Coverage here is filed against cybersecurity, application security, ai security and prompt injection. The sections sitting closest to it in the Tech Agents taxonomy are AI Agents, APIs and DevOps. Of those, AI Agents and APIs carry published work so far.
Topics in this section