Blog
Latest Posts
6 postsBy me5 min read
The Hidden Policy Inside Agent Skills
A skill is not instructions. It is policy the agent silently follows.
AI AgentsAgent SkillsMCPSecurityContext EngineeringSoftware Supply Chain
By me9 min read
The Carrying Cost of AI Agent Tools
Every installed skill and MCP server charges rent on later sessions.
AI AgentsMCPAgent SkillsContext EngineeringEvaluationDeveloper ToolsSoftware Engineering
By me8 min read
Agents Need to Forget on Purpose
Persistence is easy to add and hard to govern. Forgetting needs a policy.
AI AgentsAgent MemoryRAGEvaluationDeveloper ToolsLLM SystemsMCP
By me8 min read
Agent Workflows Need Operating Records
An agent run is only reviewable if it leaves an operating record.
AI AgentsDeveloper ToolsPlatform EngineeringEvaluationSecurityAgent Workflows
By me7 min read
Agent Systems Are Becoming Control Planes
Agent products are converging on control planes, not better chat boxes.
AI AgentsRAGEvaluationLLM ServingPlatform EngineeringDeveloper Tools
By me8 min read
What I Learned This Week About Building Reliable AI Agents
The engineering stack around the model is where reliability comes from.
AI AgentsRAGEvaluationDeveloper ToolsLLM ServingSoftware Engineering