Claude Code Finds 190K Obscene Terms Hidden in 575K Protein Sequences
leikai · reddit · 2026-10-21
A Redditor used Claude Code to scan UniProt Swiss-Prot's 575K curated protein sequences (208.5M amino acids) for Urban Dictionary slang. After filtering 3.5M terms down to 742K (stripping spaces, dropping letters B/J/O/U/X/Z, 4+ letters), 191,876 terms appear verbatim. Longest match: STINK TAINT in an insect virus protein; hits include PEEPEE and WANK in human sperm proteins. Claude Code wrote the full pipeline plus a per-position verification check.
More from coding & agent
- Building WhatsApp agents for clients: who handles the human handoff? — flavia-f · 2026-10-21
- Too many MCP tools: internal data says 77% of agent tokens are wasted — Jazzlike-Form9669 · 2026-10-21
- Agents and software companies can coexist — the bar is trust, not code quality — andreisavu · 2026-10-21
- Buying software means paying for a team that decides how it improves — can agents replace it? — andreisavu · 2026-10-21
- Claude's $200 plan equals $11k of API usage — can agent startups compete? — Victor_eu · 2026-10-21
- 90% of big-company knowledge lives in heads, and "company brain" startups are in for pain — gabriel1 · 2026-10-21