The user requests that Droid from Factory integrate "Planning and Exploration subagents" and an "AskUserQuestions tool" similar to those found in Claude Code. These features are crucial for keeping the main context cleaner, improving the overall workflow, and enhancing planning quality.
I spent the weekend testing Droid from Factory - the coding CLI that gives you Claude Code's experience across any AI model. On paper, it's perfect. The pitch: Use Codex from OpenAI for planning, Claude for frontend work, whatever model fits the task - all in one tool. No more jumping between different CLIs. Same slash commands, just pick your model. Here's why I'm sticking with Claude Code. The Feature Gap Claude Code recently added Planning and Exploration subagents. Droid doesn't have them yet. These keep my main context cleaner and make the whole workflow better. Same with the AskUserQuestions tool, which drastically improves planning quality. I rely on these features daily, and I'm not ready to give them up. The Model Reality I thought I'd want to switch between models for different tasks. Turns out, Sonnet 4.5 works well enough for everything I throw at it. Planning, frontend, debugging - I haven't found a case where switching models would be worth the context switch. Maybe I'm not pushing hard enough. Or maybe "good enough across the board" beats "perfect for each task." The Math Claude Code: $200/month flat rate. I could probably drop to $100 since I'm nowhere near the limits. Droid: Pay per token. Based on my usage patterns, I'd be looking at $400-500/month. The math makes sense if you're paying for 2-3 different tool subscriptions. But I'm not. I'm paying for one, and it covers everything I need. What Would Change My Mind If I found myself really needing different models for different tasks, the extra cost would make sense. Or if Droid pulled significantly ahead on features. But right now? I've found my workflow, and it's working. Multi-model flexibility sounds like a superpower until you realize you're paying double for options you don't use. This isn't a knock on Droid - it's a solid tool solving a real problem for people who actually need multiple models. I'm just not one of them. Yet. What's your AI coding setup? Are you using multiple models, or have you found your "good enough" sweet spot like me? #AI #DeveloperTools #ClaudeCode #CostAnalysis #BuildingInPublic