SimeonOnSecurity’s Articles

Size a 16GB local LLM setup around model weights, KV cache, and prompt processing. Compare Qwen3.8 27B memory budgets, GPU bandwidth, and ownership costs.

Choose Claude Code extensions for planning, code review, knowledge management, design, and automation. Compare ten tools, dependencies, and practical checks.

Compare nine agent skill projects by workflow, host support, dependencies, and cost. Learn how to install, inspect, and test the skills you need.

Compare Deep Agents, Claude Agent SDK, OpenAI Agents SDK, Pydantic AI, Pi, and LangGraph by tool execution, memory, recovery, and integration effort.

Compare Claude Code in the terminal and desktop Code workspace, including shared settings, session handoff, scripting, visual review, and permission choices.

Compare six terminal coding agents by workflow, automation, model access, permissions, and review effort. Build a shortlist with a repeatable repository trial.

Choose between Codex in the terminal and OpenAI’s desktop coding workspace. Compare scripting, worktrees, visual review, schedules, and execution environments.

Compare Copilot’s terminal agent with its VS Code experience, including editor context, visual diffs, agent mode, CLI integration, and account policies.

Compare graphical coding agents, separate full editors from extensions, and examine Roo Code shutdown context, model access, approvals, and review workflows.

Run a local coding agent with Strata and OpenCode. Examine reported RTX 4060 Ti results, RAM requirements, prompt reuse, reasoning budgets, and the limits of replacing a paid subscription.