Dev evals Jev on old followups, says it fully fixes agents freezing mid-task
willcb · x · 2026-09-19
Developer willcb describes a common agent failure: with followups or interactive discussion enabled, agents announce "I'll do XYZ next" then freeze without doing it. He eval'd Jev on a batch of old followups where he'd said similar things, found it quite good, and has been using it in production. The mechanism: a soft nudge message is sent only when P(yes) >= 0.5, nudging the agent to keep advancing unfinished work.
Related event: Developer tests nudge prompts to stop agents stalling early(2 posts)→
More from coding & agent
- UnCorreoTemporal MCP server lets AI agents grab OTP codes and verification links with zero human help — modelcontextprotocol · 2026-09-19
- Building 3D-printable models with GPT-6 Astra + Blender MCP, and offloading geometry to Higgsfield to save tokens — Julianw20012 · 2026-09-19
- Turned a $2 Bluetooth selfie remote into a physical push-to-talk clicker for agentic coding — _7xen_on · 2026-09-19
- What won't you let an AI agent do? For many, it's speaking in your name — Luvena21 · 2026-09-19
- willcb: in-context learning for personal continual learning needs neurosymbolic harnesses — willcb · 2026-09-19
- Scope drift: coding agents quietly modify files beyond the task at hand — Fantastic-Sleep-3352 · 2026-09-19