humanizelisted

# Humanize Rewrite prose to remove AI tells. Preserve meaning, strip the cadence. ## When to invoke User pastes text and asks to clean it up, flags something as sounding like AI or ChatGPT, asks for "tells" to be removed, or asks for general humanizing of a paragraph, email, blog post, doc, or message. ## Workflow 1. **Hard-tell scan.** Grep for deterministic markers: ```bash grep -nE 'utm_source=(chatgpt|openai|copilot)|referrer=grok\.com|:contentReference|oai_citation|turn0(search|image|news|file)|grok_render_citation|grok-card|attached_file:[0-9]|attributableIndex|【[0-9]+†|20[0-9]{2}-XX-XX' <file> ``` 2. **Triage.** Read against [references/tells.md](references/tells.md) and classify each candidate change by risk tier (see [references/risk-tiers.md](references/risk-tiers.md)): - **Tier 1** (mechanical, can't change meaning): apply silently. Em dashes, curly quotes, hard-tell markers, placeholder text. - **Tier 2** (low-risk swaps): apply, then mention in the closing summary. AI vocab swaps, didactic disclaimers, superficial tails, weasel attributions, copulative restoration. - **Tier 3** (could change tone, meaning, voice, or structure): pause and ask. 3. **Apply Tier 1 and Tier 2.** Use patterns from [references/rewrite-patterns.md](references/rewrite-patterns.md). Work category by category. 4. **Ask about Tier 3 decisions.** Use the `AskUserQuestion` tool, one decision at a time, with a recommended option labeled "(Recommended)". Wait for the an