borrowed-brain-pro is a plain-text instruction file (SKILL.md) for any AI that takes a system prompt — Claude, ChatGPT, local models. No code, no dependencies.The problem it s solving: think like X prompts universally produce motivational-poster output, because they re built from someone s polished public image. Every profile ends up sounding the same.This one works in layers: 1. Primary voice (interviews, essays, talks the person wrote themselves) 2. Documented decisions (why they made specific calls, on the record) 3. Third-party read (how colleagues and journalists describe how they think) 4. Criticism and failure — this category is required, not skippableIt extracts atomic facts with dates first, then synthesizes principles. Every principle has to trace back to something specific. If the failure section comes up empty after genuine searching, the profile says so rather than padding it.*Hard guardrails (not optional):* - No fabricated quotes. Direct quotes under 15 words, one per source. - Everything else paraphrased. - Framed as based on public material, their approach seems to be... — never stated as the person s verified private views. - Declines to profile private individuals. - For contested figures: documents reasoning style, not private motives.*9 profiles ship with it* — all built by actually running the Distill workflow with live web research, not mocked up. Includes a profiles/INDEX.md that the skill reads to suggest relevant profiles when you describe a decision without naming anyone.The repo also includes claude-ai-bundle.md — SKILL.md + all 9 profiles merged into one file you paste as a system prompt, for people using Claude.ai or ChatGPT without filesystem access.MIT licensed. Both English and Simplified Chinese READMEs.https://github.com/DOTfei/borrowed-brain-pro