Re:Ask HN: What's your current AI agentic setup?
by sivp on Tuesday November 10, 2026 @11:26AM
(#50442587)
I am using Claude for 90 percent of my work, for coding and planning tasks I use Opus 5.5, and for code, plan and security review I use Fable + Ponytail, I also added some skills for engineering, UI, system design and security installed for all users using the repo, also have a project level md files and repo level md files for Memory, Security, Claude, Agents etc.
For PR rebasing, fixing dependencies I use Sonnet, I also dispatch multiple sonnet model to probe a repo internally to find issues or before writing a plan.
I use claude code in VSCode and use Playwright MCP if I need to do some UI sanity testing, web and mobile compatibility, I run claude in safe mode.
I also use Gemini 4 Argon, as an additional reviewer if needed for documents and for an high impact tasks. Not an super good setup like some others, I am open to suggestions.
For PR rebasing, fixing dependencies I use Sonnet, I also dispatch multiple sonnet model to probe a repo internally to find issues or before writing a plan.
I use claude code in VSCode and use Playwright MCP if I need to do some UI sanity testing, web and mobile compatibility, I run claude in safe mode.
I also use Gemini 4 Argon, as an additional reviewer if needed for documents and for an high impact tasks. Not an super good setup like some others, I am open to suggestions.