✦ blog

Notes from the workshop.

Benchmarks, releases, and what we learn building a fullstack scaffolder — written up with the data attached.

June 12, 2026

ScaffBench: measuring coding agents on real fullstack scaffolding

72 runs across Claude Code and Codex CLI — six models, three creation paths, four project specs. We measured wall-clock time, output tokens, dollar cost — and whether the generated project actually installs and builds.

benchmarkmcpclaude-codeagents