Coding Agents & Language Evolution: Navigating Uncharted Waters
summary
Argues that practical fixes beat predictions: generate agents.md with new apps (100 generated Phoenix apps expose common agent mistakes), reduce reliance on Model Context Protocol (MCP) because it forces pull workflows and duplicates local capabilities, and prefer custom runtimes that map DOM to templates, run JS-driven browser tests and DB checks (demo verifies persisted email change). Highlights the agent client protocol (ACP) and the lethal trifecta — private data + external comms + untrusted content — as the real security risk.
AI-written summary · about
chapters (10)
- 0:00Intro
- 1:59Improving coding agents
- 2:53Improving agents: Instructions
- 8:54Improving agents: Tools
- 18:15Improving agents: Runtime
- 18:42Demo
- 24:17Improving agents: Runtime continued
- 31:31Building integrated runtimes
- 36:37Summary
- 37:07Outro
related talks
32mProgramming language evolution in the AI era
33mAgent Experience
40mHow To Tame Your Agents?
48mAI Coding Agents and How to Code Them
40mAI Coding Agents and how to code them
01h 05mPanel: Agentic SE & Open Source in the FMware Era – Toward Trustworthy, Context-Aware FMware
01h 03mAI Coding Agents and how to code them
34mProgramming Languages for AI Programing Agents