Agent Harness

9 stories tagged Agent Harness.

Agent Harnesses, Not Bigger Context Windows, Decide Long Task Success

A new analysis argues agent harnesses beat raw context windows on long tasks. We examine the four mechanisms, the missing benchmarks, and what it means.

HarnessDev Asks Whether AI Models Can Engineer Their Own Tools

Yuki Okonkwo1 month ago

AI. Yuki Okonkwo1 month ago

HarnessDev Asks Whether AI Models Can Engineer Their Own Tools

HarnessDev from ByteDance Seed and partners tests whether LLMs can build their own agent harnesses. The catch: only 34 of 64 changes generalized. Here's what that means.

Pi Coding Agent Gets Subagents and MCP Through Two Extensions

Yuki Okonkwo1 month ago

AI. Yuki Okonkwo1 month ago

Pi Coding Agent Gets Subagents and MCP Through Two Extensions

A new walkthrough shows how the minimal Pi coding agent gains parallel subagents and MCP support, plus what the numbers reveal about context and cost.

Man smiling next to whiteboard diagram explaining prompt caching architecture with system prompts, tools, and pricing tiers

How Prompt Caching Cuts AI Agent Costs

Yuki Okonkwo2 months ago

AI. Yuki Okonkwo2 months ago

How Prompt Caching Cuts AI Agent Costs

Prompt caching can dramatically reduce AI agent costs—but only if your setup preserves reusable prefixes. Here's what actually gets cached and what kills it.

Overhead view of hands typing on laptop displaying "The Factory: Multiplexing Engineering Yield" with yellow "SR. ENG" text…

Agentic Engineering: The 5-Pillar Framework for 2026

Dev Kapoor5 months ago

AI. Dev Kapoor5 months ago

Agentic Engineering: The 5-Pillar Framework for 2026

IndyDevDan's 5-pillar agentic engineering framework is technically sharp—but who actually gets to build these systems, and at what cost to the OSS tools underneath?

Man in Argentina jersey gestures while discussing chaos in agent town, with crime statistics displayed on screen beside him

What AI Town Experiments Actually Teach Us About Agents

Yuki Okonkwo5 months ago

AI. Yuki Okonkwo5 months ago

What AI Town Experiments Actually Teach Us About Agents

Emergence AI's 15-day virtual town experiment revealed wildly different AI behaviors—and the real lesson has nothing to do with which model is "best."

Bold text reading "Code Scaler" in yellow box with 3D cube icon surrounded by four outward-pointing arrows on dark background

Claude Code at Scale: The Harness Is the Product

Yuki Okonkwo5 months ago

AI. Yuki Okonkwo5 months ago

Claude Code at Scale: The Harness Is the Product

Claude Code works fine for small projects. But at scale, the model matters less than the harness around it. Here's what that actually means in practice.

Two men in a tech presentation setting with AI workflow diagrams behind them, discussing long-running agents and persistent…

AI Agents Running for Hours—and Who's Accountable

Samira Barnes5 months ago

AI. Samira Barnes5 months ago

AI Agents Running for Hours—and Who's Accountable

Anthropic's Prabaker and Wilson reveal the engineering behind long-running AI agents—and raise accountability questions regulators haven't caught up to yet.

Bold white text reading "SELF IMPROVING AGENTS" on a black background with a vibrant magenta-to-cyan gradient border

AI Agents That Optimize Themselves While You Sleep

Dev Kapoor6 months ago

AI. Dev Kapoor6 months ago

AI Agents That Optimize Themselves While You Sleep

Kevin Guo's AutoAgent extends Karpathy's auto-research loop to let AI agents rewrite their own operational code overnight. What happens when agents program agents?