My Actual Tool Stack
No affiliate links. No sponsorships. Just what I use every day.
People ask me what tools I use. Here is the honest answer with no marketing spin.
For Agent-Driven Development
- Claude Code. My go-to for autonomous dev work. Point it at a codebase, give it a task, it executes. Reads files, writes code, runs tests. The key is the system prompt and the context you feed it.
- Cursor. Primary code editor. Agent-driven. I use it when I need tighter control, editing specific files, reviewing diffs in real time. Claude Code for autonomous runs, Cursor for guided work. Different tools for different levels of oversight.
For Always-On Agents
- Open Claw. The runtime that keeps my agents running. Think of it as the always-on execution layer. Each agent gets its own instance with its own context. They do not share state unless I explicitly bridge them.
For Prompt Management
- Custom prompt library. I built a simple app for managing my prompts. I have dozens of operational playbooks. Every prompt follows the same structure: role, objective, constraints, workflow phases, output format, hard constraints. I tried existing tools. None of them worked the way I needed.
For Everything Else
- Claude. Deep technical work and strategic thinking.
- ChatGPT. General questions and ideation.
- Todoist. Task management and household workflow coordination.
- Telegram. Delivery channel for agent summaries.
Always remember the tools matter less than how you use them. Regardless if you are building agents or houses :-)
I will walk you through each tool and help you pick the right ones for your workflow.