andyed

Prose Audit — Security skill for Claude Code

Security community

Lint paper drafts for AI-tell prose patterns.

How to install Prose Audit

Installs to ~/.claude/skills/andyed-science-agent-prose-audit/SKILL.md

Terminal

mkdir -p ~/.claude/skills/andyed-science-agent-prose-audit && curl -fsSL https://raw.githubusercontent.com/andyed/science-agent/HEAD/commands/prose-audit.md -o ~/.claude/skills/andyed-science-agent-prose-audit/SKILL.md

Restart Claude Code, or start a new session, for it to be picked up.

What Prose Audit does

description: Lint paper drafts for AI-tell prose patterns. Args: [--severity=warn] [--no-pencil] [--no-muriel] [--no-native] [--summary] [--json]

Run the science-agent prose audit with the user's arguments:

NODE_PATH="${CLAUDE_PLUGIN_DATA}/node_modules" node "${CLAUDE_PLUGIN_ROOT}/cli.js" prose-audit $ARGUMENTS

Then summarize. Lead with the highest-severity rule clusters (where the same rule fires many times — that's the actionable signal). If `$ARGUMENTS` is