Skip to content

Security: Fernado03/zoo-flow

Security

SECURITY.md

Security policy

Zoo Flow is a configuration template for Zoo Code. It does not ship runtime services, but it does shape how an AI assistant behaves inside your workspace. Treat the modes, rules, and commands as part of your security surface.

Reporting a vulnerability

If you believe you have found a security issue in this template — for example, a rule, command, or skill that could be used to exfiltrate secrets, escalate file-system access beyond the documented bounds, or bypass the path safety guarantees — please report it privately.

  • Open a GitHub Security Advisory on the repository, or
  • Email the maintainers listed on the repository's GitHub profile.

Do not file a public issue for security reports. Once we have a fix, we will coordinate disclosure with you in the advisory.

What is in scope

  • The mode definitions in templates/full/.roomodes.
  • The always-on rules under templates/full/.roo/rules/.
  • Mode-scoped rules under templates/full/.roo/rules-*.
  • The slash commands under templates/full/.roo/commands/.
  • The skills under templates/full/.roo/skills/.

What is out of scope

  • Vulnerabilities in Zoo Code, the underlying model providers, or third-party tools the template references. Report those upstream.
  • Issues that depend on a user manually disabling path-safety rules or removing the orchestrator's routing matrix.

Handling secrets

This template never asks you to commit secrets. The .gitignore excludes .env, .env.*, and local journals. Several skills explicitly require redaction of API keys, credentials, and PII before producing output. If you find a place in the template that pulls real credentials into context, treat that as a security bug.

There aren't any published security advisories