Cross-Model Code Review: Having Claude and Copilot CLI Battle Over Refactoring
DanWahlin · x · 2026-09-05
Developer Dan Wahlin shares a workflow: running Claude and Copilot CLI simultaneously to clean up code and having them review each other's changes. The result is both consensus and "that guy's crazy!" pushback between the two agents.
He endorses cross-harness/cross-model reviews — letting different models critique each other surfaces blind spots that a single model misses.
More from coding & agent
- Unvalidated JSON, not bad reasoning, is why AI agents crash in production — samuelcolvin · 2026-09-05
- Astra xHigh Follow-Up: Codex vs AGNT Side-by-Side Comparison Opens the Floor — NathanWilbanks_ · 2026-09-05
- Same Prompt, Different Harness: Astra xHigh Yields Notably Different Results in Codex vs AGNT — NathanWilbanks_ · 2026-09-05
- Same prompt, same seed image: the coding harness still matters — NathanWilbanks_ · 2026-09-05
- Opinion: models grinding for days against an uncheatable verifier is the true capability frontier — mike64_t · 2026-09-05
- Self-improving agents: experience should change the whole agent, not just its memory — furongh · 2026-09-05