Articles

• How to Harden Claude Code Against Prompt Injection April 1, 2026

  Three layers of defense for Claude Code: out-of-the-box protections, CLAUDE.md rules, and hooks that block exfiltration before it executes.

• How I Built a Research Agent That Doesn't Hallucinate March 28, 2026

  I turned Anthropic's hallucination reduction guide into a working research agent. Here's the architecture, the guardrails, and a real example.