Contextia szerző: Contextia

Contextia is a local, privacy-first guard that flags secrets in the chat composer before they ever leave your machine.

Developers paste code, logs and configs into AI chats every day — and those often contain live API keys, tokens, private keys and connection strings. Contextia catches them at the moment of typing or pasting, entirely on your device.

What it does: • Detects 40+ credential types — AWS, GitHub, GitLab, Stripe, Slack, OpenAI, Anthropic, Google, Azure, Twilio, SendGrid and more — plus PEM private keys, .env secrets, database connection strings, JWTs, and personal data like Luhn-valid credit-card numbers and IBANs. • Explains why each match was flagged, so it's never a mystery. • Underlines the offending text and shows a popover: Redact, Allow once, Allow all, or Allow this pattern always. • Add your own data to always redact (custom values and regex patterns). • Modes: Warn, Auto-redact before sending, Block until resolved, or Off. • A local detections log and stats — stored on your device, never the secret value.

Privacy by construction: • Zero network requests. No accounts, no servers, no telemetry, no third parties. • Runs only on chatgpt.com and claude.ai. Requests no other access. • Open source (MIT) — audit every line: https://github.com/sbr0nch/contextia

Contextia is a safety net, not a guarantee: rule-based detection can miss things, so treat it as a guardrail, not proof that a paste is clean.