One Prompt Fix Stops GPT-6 Coding Agents From Pausing on Casual Chatter
altryne · x · 2026-09-25
altryne shared a prompt snippet for agents.md that stops coding agents from pausing all work when humans send casual messages.
The core idea: instruct the model that human messages are not always steering — only messages that start, change, cancel, or continue work count as directives. Casual chat, acknowledgements, or questions should be answered normally from current context (or with a single reaction), without reading, writing, restoring/clearing checkpoints, updating Up Next, or scheduling waits. Directly copy-pasteable for teams running long tasks on GPT-6-class models.
More from coding & agent
- Climbing analysis in 3D with iPhone LiDAR: open-source SAM 3.1 + ViTPose demos — dosco · 2026-09-25
- Diffusion LLM goes production: Augment Code's Mercury 2.5 switch cuts latency 82%, cost 90% — cen6wkf · 2026-09-25
- Researcher: Time Series Will Hand Your Ass to You, Even in Agent Orchestration — RexDouglass · 2026-09-25
- Indie creator shares a storyboard workflow for AI-generated action sequences — Educational_Wash_448 · 2026-09-25
- AI Search Startup Octen Promotes $1 Per 1,000 Web Search Calls for Agents — kimmonismus · 2026-09-25
- IterSynth: Role-Decoupled Search Agent Beats Prior 8B Peers by 4.2% — ZJU-REAL · 2026-09-25