Running Multiple AI Agents for Planning and Reviews: What's Your Approach?

Hey HN,

I’ve been experimenting with coding agents for about a year. I paid for Cursor and liked it until I hit the monthly limit, then tried Claude Code and now use it most. I also keep a ChatGPT subscription because I created a couple of Custom GPTs that others still use. Codex is now part of the ChatGPT plan, and Gemini CLI provides 1,000 free requests per day.

I suspect I’m not getting full value from all these tools.

How do you handle multiple providers?

I use them for brainstorming and switching personas. For planning, I ask Claude Code to spawn subagents (Gemini, Codex, Cursor). I run other coding CLIs via Roundtable MCP https://github.com/askbudi/roundtable/ , assign each one planning tasks, and then merge their plans. For landing pages, I send subagents to critique and iterate. The mix of models and system prompts helps me see more angles.

Curious to hear how you use them.