Skip to content

Security: Zizka-ai/ZizkaDB

SECURITY.md

Security Policy

ZizkaDB is an operational database for AI agents. It stores event history, API keys, user accounts, and embeddings metadata. We take security reports seriously.

Supported versions

Version Supported
main branch (latest) Yes
Latest GitHub release Yes
Older releases Best effort — upgrade recommended

Managed cloud at db.zizka.ai runs the latest stable deployment from this repository.

Reporting a vulnerability

Please do not open a public GitHub issue for security vulnerabilities.

Email founder@zizka.ai with subject line ZizkaDB Security and include:

  1. Description of the issue and potential impact
  2. Steps to reproduce (proof-of-concept if available)
  3. Affected component (API, dashboard, SDK, MCP, self-host Docker stack)
  4. Your environment (self-hosted vs managed, version/commit if known)
  5. Whether you would like public credit (name/handle) when fixed

We aim to acknowledge reports within 3 business days and provide a status update within 7 business days.

What to report

Examples of in-scope issues:

  • Cross-tenant data access (viewing another user's events, agents, or API keys)
  • Authentication bypass (OTP, JWT, dev key misuse in production)
  • API key leakage or weak hashing
  • SQL injection or unsafe query construction
  • Remote code execution in API, dashboard, or MCP server
  • SSRF via dashboard or API proxies
  • Privilege escalation (user → admin, tenant → tenant)
  • Insecure defaults on production deployments (DEV_API_KEY exposed, open admin routes)

Out of scope (please use regular issues instead):

  • Social engineering
  • Denial of service without a practical exploit path
  • Issues in third-party services (Stripe, email providers) outside our control
  • Missing security headers with no demonstrated exploit
  • Vulnerabilities in dependencies already fixed in main

Safe harbor

We will not pursue legal action against researchers who:

  • Make a good-faith effort to avoid privacy violations and service disruption
  • Do not access or modify data belonging to other users
  • Report findings promptly and allow reasonable time to fix before public disclosure

Disclosure timeline

  • We prefer coordinated disclosure
  • We will work with you on a reasonable embargo (typically 90 days or until a fix is released)
  • After a fix, we may publish a security advisory and credit reporters who opt in

Production hygiene (self-hosters)

If you run ZizkaDB yourself:

  • Set ENV=production and unset DEV_API_KEY
  • Configure EMAIL_* for OTP; use TLS in front of the API and dashboard
  • Never run docker compose down -v on servers with real data
  • Back up Postgres regularly (infra/backup-postgres.sh)
  • Keep Docker images and dependencies updated

See Production Deployment wiki.

Contact

founder@zizka.ai — security reports and code of conduct enforcement

There aren't any published security advisories