Split frontend and backend coding between Claude and Codex
A developer uses Claude Design for front-end work and Codex for backend work, after finding OpenClaw and Hermes fell apart on real work. ·
Done withClaudeCodexOpenClawHermes Agent
- What they did
- The commenter pays for two Claude Ultra and two ChatGPT/Codex Pro accounts, plus a few local models for small jobs like text formatting and TTS. They use the default CLI and desktop apps and built their own tool to combine them. For the past month, Claude Design handles front-end work, given plenty of references, and Codex handles backend work.
- How it went
- Codex reliably finishes backend work but designs poorly. Claude designs well but struggles to bring anything to completion. OpenClaw and Hermes fell apart on real work, so they gave up on them.
- Worth knowing
- Claude's design quality depends on supplying enough references, and running multiple top-tier subscriptions across both vendors is a real recurring cost.
Try it yourself with Claude
In [project folder], build the backend for [describe the feature], including [API endpoints, database, or auth needs]. Follow the existing code conventions, run the tests, and write a short note listing the endpoints and data shapes the frontend will need. You are done when the tests pass and that note is saved as [file name].
Want help getting this working? Join the Vibe Coding Academy ↗ Run by Alex Finn, who also runs this site.
Source: HN · Nov 4, 2026
Related use cases
Build a C compiler with a team of parallel agents
Anthropic's engineering write-up on tasking Opus 4.6 agent teams with building a C compiler.
Search and apply to jobs on LinkedIn
OpenClaw with browser access made a resume, searched listings and applied to 100+ jobs in three days, but also DMed people on LinkedIn without consent.
Handle a personal mailbox daily with a local model
The poster runs Hermes with free local models via Ollama on their own GPU, using a daily cron job to process their personal mailbox.
Review Linux kernel patches
Google engineers launched Sashiko, an agentic AI code reviewer for the Linux kernel.
Custom agent
Build a full CRM with email, video and calendar
A community member used Replit Agent over a couple of weeks to build a CRM with email integration, video sales letter recording and meeting scheduling.
Check code against user intent with review agents and tests
A developer has agents re-read all user messages to re-derive intent, then checks code and tests against it, tracking work in a reviewed todo file.
Custom agent
Get 5 like this every morning
The best things people got an AI agent to do, each with the prompt to try it.