This project is not yet published as a stable package. Security fixes should target main.

Open a private security advisory on GitHub or contact the repository maintainers through the published project contact channel. Do not include secrets, live customer data, call recordings, or full transcripts in a public issue.

• Never commit .env files.
• Never commit SQLite databases, call recordings, transcripts with personal data, or production logs.
• Keep API keys, telephony credentials, Clerk keys, and database passwords in environment variables or a secret manager.
• Rotate any credential that was committed or printed in public logs.

For DEPLOYMENT_MODE=hosted, keep these controls enabled:

• REQUIRE_DASHBOARD_TOKEN=true
• REQUIRE_WEBHOOK_SIGNATURES=true
• AGENT_DASHBOARD_TOKEN_SECRET and AGENT_MEDIA_STREAM_TOKEN_SECRET set to separate 32-plus character random values
• DASHBOARD_ALLOWED_ORIGINS restricted to the hosted dashboard origins
• INBOUND_ORG_ROUTES configured for every hosted inbound telephony entry point
• Stripe webhook signature verification enabled with STRIPE_WEBHOOK_SECRET
• Usage ingest protected with BILLING_USAGE_INGEST_SECRET, X-Usage-Timestamp, and X-Usage-Signature
• /health/details reachable only with a dashboard token that includes calls:read

The agent fails startup in hosted mode if dashboard tokens, media stream tokens, webhook signatures, usage ingest, or inbound route mapping are unsafe.

Do not expose secrets through NEXT_PUBLIC_* variables. Public variables may include app URLs and publishable browser keys only.

Voice agents can process sensitive calls. For production use, review consent, recording, retention, access controls, emergency routing, and compliance requirements for your jurisdiction and industry.