Get a real second opinion: send the failure, not the agent's theory
Ok_Negotiation_2587 · reddit · 2026-11-05
Pasting your coding agent's diagnosis into another chatbot mostly gets the same diagnosis back, because the summary carries a theory the second model just refines — a false confirmation.
The author's workflow: start a fresh chat with only the failing behavior (expected/actual/reproduce command/pre-diff code), ask for the three most likely causes ranked plus one check to rule out each — no fixes. Don't include the agent's diff, since a diff is also a theory. Only then introduce the agent's theory and ask what evidence supports or contradicts it; hand the checks back to the agent and make it paste real output.
He suggests A/B testing this on your next stuck run. Also promotes his AI Toolbox browser extension for saved prompts and cross-chat error search.
More from coding & agent
- Adversarial speculative decoding boosts long-horizon agent accuracy, paper finds — SubconsciousDev · 2026-11-05
- Axiom ships 16M-line AI formalization of the Stacks Project, countering 'AI code is slop' — Scobleizer · 2026-11-05
- Oumi launches FAiE, an agent that builds and trains your specialized AI agents in 15 minutes — JaynitMakwana · 2026-11-05
- Five ways an agent run can end: success, giving up, budget, human handoff, failure — ClickOk5811 · 2026-11-05
- Solo dev builds Suno + three.js music video pipeline with Claude Code in hours — Glass-Garden-5888 · 2026-11-05
- An obscure git command yields a 12% speedup, tested on the Linux kernel repo — DanielLockyer · 2026-11-05