posts
Articles tagged under posts.
current limitations·
The Hyper-Reactive Architect - When Agents Can't Sit Still
Why AI agents pivot to destructive new paths after a single glitch and how to build in patience.
guidelines·
AGENTS.md / CLAUDE.md - The way I like to work with AI Agents
Its not just about the prompt, its about the workflow and the current state of the project.
patterns·
Multi-Agent Orchestration - From Scripts to Swarms
Designing robust handoff patterns and state management for multi-agent systems.
current limitations·
Presumptive Inference - The Silent Agent Killer
How agents assumes success and how to architect guardrails to prevent it.
current limitations·
Why Agents Prefer Shortcuts - And Why You Must Stop It
Understanding why LLMs prefer fragile scripts over robust systems and how to enforce Agentic Engineering.
internals·
Observability in Agentic Flows - Debugging the "Black Box"
How to instrument agentic systems for deep traceability and debugging.