Sveriges mest populära poddar

Your Coding Agent Passed the Benchmark—Then Failed the Refactor

19 min•25 augusti 2026

Om avsnittet

Most coding-agent benchmarks reward contained tasks, but real repositories demand changes across boundaries, tests, migrations, and documentation. Fictional AI hosts Alex and Sam show how to run a five-part refactor trial that exposes whether an agent can preserve architecture—not merely produce a passing patch.

Fler avsnitt

Claude Code Cast med AI World finns tillgänglig på flera plattformar. Informationen på denna sida kommer från offentliga podd-flöden.

Your Coding Agent Passed the Benchmark—Then Failed the Refactor

Om avsnittet

Fler avsnitt

An Agent Missing One Key Burned $100. Claude Code Projects Needs This Stop Rule

Claude Code's Cache Fixes: Check Before You Switch Models

OpenAI Cut Off Cursor. Five Days Later, Four Models Went Down.

Same Model, 70x the Tokens—Your Harness Sets the Bill

Passing Tests Isn't Enough for Your Next Coding Agent

Your OpenClaw Updates Need a Canary, Not Courage

Claude Code Changed Engines—Your Evals Just Broke

Better Agent Tools Made Code Review Worse

Your AI Coding Benchmarks Are Lying To You

The Tiny Local Model That Changes Your Agent Budget