A three-model coding pipeline: Astra for GitHub issues, rival models to code and review, Sol for final testing
craigbalding · x · 2026-09-19
craigbalding shares a "high IQ sandwich" multi-model coding workflow:
- Use Astra to generate detailed GitHub issues — but don't burn your codex quota on this; use ChatGPT set to Pro 6 instead
- Have competent models do the coding AND the review, using different model families and a clean session per issue
- Escalate to Sol for review escalations and final testing ("Boss Mode")
The core idea: let different models check each other's work while preserving quota for the expensive steps.
More from coding & agent
- Fully Local, Private Computer-Use AI Arrives — Runs on a 3060 Ti 12GB — WolframRvnwlf · 2026-09-19
- ActObs: Supervising Observations During SFT Boosts Agent RL by up to 4.2pp — kastnerkyle · 2026-09-19
- Cloudflare open-sources agentic-inbox, a self-hosted AI email client running entirely on Workers — dinasaur_404 · 2026-09-19
- Pi hits Pareto frontier on both benchmarks with just four tools: read, write, edit, bash — arena · 2026-09-19
- "Let me check in with my agents" is now a daily phrase in this developer's marriage — josh_bickett · 2026-09-19
- "I gave jev eyes": devs celebrate gloriously pointless AI experiments — cocktailpeanut · 2026-09-19