New Codex plugin optimizes long-task polling, reducing token waste
ssh4net · x · 2026-08-15
A developer released the Codex MCP Longrun plugin to address polling inefficiencies when AI coding agents handle long-running tasks like builds or tests. This local STDIO MCP server uses an asynchronous workflow to return a job ID immediately, avoiding blocking calls and model-visible wait loops. This significantly reduces context usage and token waste. The project is currently in a Linux/WSL pilot phase.
More from coding & agent
- Open-Source Course: Build a Second Brain AI Assistant with Agents and RAG — tom_doerr · 2026-08-15
- Sorting Agent Arena by 'Praise vs Complaint' aligns rankings with user preferences — MilesCranmer · 2026-08-15
- Pydantic AI Harness v0.21.0 Released with Coder and Researcher Harnesses — samuelcolvin · 2026-08-15
- Vibe-Coding Tip: Have AI Auto-Generate and Update Backend Flowcharts — eptwts · 2026-08-15
- 3 AI Agent Patterns Explained: Harness, Loop, Graph — Accomplished_Job_76 · 2026-08-15
- Grok admits password access risk; agent browser security under scrutiny — Imaginary_Dinner2710 · 2026-08-15