Skip to content
ai101.tools
navigateopenescclose
Claude+372Whisper+228LangChain+168Codex+223NotebookLM+276DALL-E 3+192DeepL+249n8n+208Topaz Video AI+153LlamaIndex+161
Verification Before Completion logo
Skill

Verification Before Completion

Blocks any completion claim until a command has been run and its output read.

0
SaveVisit website ↗

Install

/plugin install superpowers@claude-plugins-official
Triggers on

Stands in front of every "done", "fixed" or "passing" claim, and of the commit or pull request that would follow it.

Author
Jesse Vincent (obra)
License
MIT

Before the agent says something works, it has to name the command that would prove it, run that command fresh, and read the output in full. Confidence is not evidence, and hedged wording like "should work" does not clear the gate either. The skill also catches the near-misses that usually slip through: a clean linter is not a successful build, and a subagent's report is not verified until someone checks it.

Comments(0)

Sign in to comment

No comments yet — be the first.

Similar skills

Report this comment

Why are you reporting this?