exobrain-auditlisted

# Exobrain Audit A three-phase audit that checks repo privacy, analyzes harness architecture, and researches the latest AI productivity ideas. Each phase runs semi-independently via subagents, so they can execute in parallel where possible. ## Phase 1: Privacy & Legibility Audit This is the most critical phase. The Exobrain harness repo is designed to be publicly sharable and replicable. That means **no personal data in tracked files** AND **the repo must make sense to a stranger cloning it for the first time**. ### What to scan for Run `git ls-files` to get all tracked files, then scan them for: 1. **Other people's real names or identifying info** -- full names, phone numbers, email addresses, physical addresses, Discord-to-real-name mappings, transcript speaker corrections. Generic placeholders like `[Name]` or `[Friend]` are fine. 2. **Alex's private info** -- salary, home address/ZIP, health data values, relationship details beyond what's needed for function (cycle data values, mood scores, etc.) 3. **Name-to-identity mappings** -- any table, dict, or config that maps usernames/handles to real people 4. **API keys, tokens, credentials** -- anything that looks like a secret (long random strings, `sk-`, `token:`, `Bearer`, etc.) 5. **Personal data logs** -- mood scores, cycle data entries, event preferences, message content, processing log entries with personal details ### Legibility check Beyond privacy, also check that the repo is **legible to someone who isn't Al