Compound Engineering 3.20 splits AI coding across multiple models and adds handoff snapshots
danshipper · x · 2026-07-24
Compound Engineering 3.20 pushes a multi-model coding workflow: plan in one model, implement in another, and review with a third, without manually moving context.
What changed
- Route each stage to a different model or harness with commands like /ce-plan use fable or /lfg implement with codex.
- Use oracle to have multiple models independently inspect a repo, debate differences, and recommend the best approach.
- Code and doc reviews can now automatically use a rival provider to catch issues the primary model missed.
- /ce-handoff creates a resumable snapshot of goals, constraints, dead ends, local state, and the exact command to continue.
- /ce-babysit-pr watches CI and review feedback.
The pitch is simple: the best AI coding setup is no longer one model for everything, but the right model for each job.
Related event: Compound Engineering v3.20 Introduces Multi-Model Collaboration(3 posts)→
More from coding & agent
- Builder rebuilds a Claude Code software factory around planning, implementation, review — blaizedsouza · 2026-07-24
- Default Codex CLI with GPT-5.5 scores 92.3% on XBOW, sparking benchmark fatigue — moyix · 2026-07-24
- Sebastian Raschka will discuss DeepSeek-V4, GLM-5.2 and open-weight coding agents — hugobowne · 2026-07-24
- Antigravity CLI 1.1.6 makes custom agents editable as Markdown files — rseroter · 2026-07-24
- FinanceComplexQA adds a 2,026-task benchmark for agentic reasoning on financial docs — Beihang · 2026-07-24
- Microsoft Research’s ReOPD reuses teacher prefixes to distill multi-turn agents offline — MicrosoftResearch · 2026-07-24