Scan a blob of code, text, or a unified diff for LEAKED SECRETS before you commit, push, open a PR, or paste it somewhere. Detects provider API keys (AWS, GitHub, OpenAI, Anthropic, Stripe, Google, Slack, Twilio, SendGrid, npm, Telegram, Discord, Shopify, Cloudflare and more), generic tokens, private keys (RSA/EC/DSA/OpenSSH/PGP), JWTs, database connection strings with passwords, basic-auth URLs, and high-entropy strings that look like credentials. Returns a CLEAN / REVIEW / LEAK verdict with each finding's secret type, provider, severity, line:column, a MASKED excerpt (never the full secret), and a remediation note. Use this on every diff/file you are about to share. The scan is fully local — the secret is never sent anywhere.