https://minh2416294.github.io/posts/ Recent content in Posts on Minh's Blog https://minh2416294.github.io/images/og-default.png https://minh2416294.github.io/images/og-default.png Hugo en-us Tue, 30 Jun 2026 00:00:00 +0000 https://minh2416294.github.io/posts/i-build-the-guardrails-around-ai-agents/ Tue, 30 Jun 2026 00:00:00 +0000 https://minh2416294.github.io/posts/i-build-the-guardrails-around-ai-agents/ A self-intro. I work on the controls around AI agents - what they can do, what stays human, and how the system fails when something slips. Here’s how I think, shown through the work, including its limits. https://minh2416294.github.io/posts/prompts-ask-hooks-enforce/ Sun, 21 Jun 2026 00:00:00 +0000 https://minh2416294.github.io/posts/prompts-ask-hooks-enforce/ A rule stated in the markdown is typically followed, but when it’s not, that rule belongs in code that runs before the tool itself. https://minh2416294.github.io/posts/two-bugs-in-my-own-guardrails/ Wed, 10 Jun 2026 00:00:00 +0000 https://minh2416294.github.io/posts/two-bugs-in-my-own-guardrails/ A guardrail that blocked the workflow it was meant to protect, and a permission rule that was silently dead. Both were bugs in my own config. https://minh2416294.github.io/posts/your-agent-has-no-ground-truth/ Sun, 07 Jun 2026 00:00:00 +0000 https://minh2416294.github.io/posts/your-agent-has-no-ground-truth/ Context degradation, silent failure, and miscalibrated escalation look like three separate problems. They’re all the same instrumentation gap. https://minh2416294.github.io/posts/llm-extraction-failure-modes/ Tue, 12 May 2026 00:00:00 +0000 https://minh2416294.github.io/posts/llm-extraction-failure-modes/ Every prompt engineering technique exists because a specific failure mode forced it. Here’s the failure taxonomy, not the technique list. https://minh2416294.github.io/posts/the-claude-code-config-stack/ Mon, 20 Apr 2026 00:00:00 +0000 https://minh2416294.github.io/posts/the-claude-code-config-stack/ CLAUDE.md, path rules, skills, hooks, and headless CI each have their own post. Nobody writes about how they compose or what happens when the stack drifts. https://minh2416294.github.io/posts/how-to-debug-tool-misrouting/ Fri, 10 Apr 2026 00:00:00 +0000 https://minh2416294.github.io/posts/how-to-debug-tool-misrouting/ Everyone says ‘fix your tool descriptions.’ Nobody shows how to diagnose which specific failure caused the misroute. https://minh2416294.github.io/posts/the-agentic-loop-is-not-the-hard-part/ Sun, 05 Apr 2026 00:00:00 +0000 https://minh2416294.github.io/posts/the-agentic-loop-is-not-the-hard-part/ stop_reason is six lines of code. The real engineering in agentic systems is protecting what goes into the model’s context window. https://minh2416294.github.io/posts/your-rag-eval-is-lying/ Wed, 01 Apr 2026 00:00:00 +0000 https://minh2416294.github.io/posts/your-rag-eval-is-lying/ A 0.91 faithfulness score doesn’t mean your RAG pipeline works. Most eval panels can’t see the layer that’s actually broken. https://minh2416294.github.io/posts/reach-for-an-agent-last/ Sat, 21 Mar 2026 00:00:00 +0000 https://minh2416294.github.io/posts/reach-for-an-agent-last/ A decision framework for picking LLM architecture by asking what failure costs first and why agents are the right answer less often than you think.