qdoctor

What problem does it solve? Scheduled and headless Claude Code runs often fail silently because a repo's configuration is missing pieces the user's other repos already have — most critically a Write permission entry, without which a headless run dies at its first write after doing all the work. This Skill compares the current repo's CLAUDE.md, slash commands, hooks, and permissions against the user's own other repositories to surface concrete configuration gaps. ## Core Features & Use Cases - Cross-repo configuration audit: Compares this repo's .claude/settings.json and settings.local.json against every other repo discovered from transcript cwd records, never against a generic best-practice list. - Write-permission check: Flags when permissions.allow lacks a Write or Edit entry, the most common cause of cron jobs that run for weeks and ship nothing. - Evidence-graded findings: Reports a gap only when at least half of four or more other repos have the item; otherwise it reports a note, keeping evidence separate from hunches. - Use Case: A user's nightly scheduled task produces no output. Running the audit reveals eleven of their sixteen repos allow Write in settings while this one does not, and the Skill offers to draft the fix on a branch. ## Quick Start Ask the assistant to audit this repository's Claude Code setup against my other projects and tell me why my scheduled runs might be failing.