Observed arrival · 2026-09-10
PlaybookDiff audits the instructions AI coding agents actually receive
An open-source, read-only CLI and GitHub Action compares Claude Code and Codex repository configuration.
○Open
⊠Login
$Paid
†Ads
✦Pretty
●Pro
◎Niche
◉Human
⚑Risk
ƒJS
Field notes
The analyzer accepts a repository, launch directory, and target path, then traces which instruction files and skills each coding agent receives at that location. Its displayed example follows apps/api/file.ts and reports two medium findings: a nested instruction Claude Code receives but Codex does not, plus an api-skill available only through Claude Code. The page also documents a GitHub Action, security model, limitations, and a supported-semantics matrix.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
○OpenPublic substance visible
✦PrettyNotable craft visible
●ProPolished or operationally mature
◎NicheUnusually specific use
One card from the complete issue