Remediating the eMudhra Certificate Authority CompromiseThe number 4️⃣ strikes again! What users can do about it.5 min read·kevinhwang.dev2AAOct 23
Classifying Malicious Repositories With AI — And Doing It SafelyPhishing repos begone!7 min read·kevinhwang.dev41JSep 2
The Summer AI Agents Started Going RogueTurns out AI is lazy and dangerous like us...8 min read·kevinhwang.dev65MMKJul 23
Reflections On Trusting Trust — Of Our AI ModelsWhat "local AI" misses — could that "open source" model be backdoored? 🧐10 min read·kevinhwang.dev21AJul 10
The Unreasonable Effectiveness of CLIs > MCPWhy CLI Tools + Skills often still beat MCP8 min read·kevinhwang.devJul 8
