groundtruth
01Plugin Claude Code
Bundles 7 commands, 5 hooks · 167 tokens together
Audits whether an agent actually did what it was asked. Catches the false 'Done' on Stop: missed subtasks, stubbed/no-op work, false 'tests pass', a method silently dropped in a refactor (dangling reference), and — uniquely — a standing project rule that was in context and overridden anyway. Runs on Stop, pre-commit…
not rated 7 3mo ago A
tokens not measured
original MIT