Ask your model what's wrong with your harness — an easy trick to catch silent bugs
insomniating · reddit · 2026-11-05
A developer building a custom harness shares a trick: ask the model itself what's wrong with the harness and what could be improved. Because the model has full context on the harness internals, its self-critique surfaces issues developers overlook — the author found several silent bugs this way. The approach generalizes to debugging agent pipelines and eval frameworks.
More from coding & agent
- Selmem experiments show divergent memories don't yield divergent behavior without a self-model — Tough_Conflict_8852 · 2026-11-05
- Multi-Turn Agent RL Gains Hinge on Interaction Budgets, CASIA Study Finds — CASIA · 2026-11-05
- Self-taught dev ports local voice assistant ARYA from n8n to LangGraph, opens architecture for review — Alternative_Yak_1367 · 2026-11-05
- GitHub Next's PDD Kit uses formal methods to prove AI-generated web app code is correct — 0xkarasy · 2026-11-05
- Open-Source hunch Distills LLM Judgments Onto CPU: Millisecond Decisions Under $0.03/Million Tokens — aigclink · 2026-11-05
- Meta's JiTTesting deploys LLM-generated tests to catch bugs pre-merge, cutting human review 70% — AxSaucedo · 2026-11-05