Skip to content

Security: SonyCSLParis/cortex

Security

SECURITY.md

Security Policy

Supported versions

Cortex is currently developed from its latest public main branch. Security fixes are applied there; older snapshots are not maintained as supported release lines.

Reporting a vulnerability

Please do not disclose vulnerabilities, credential exposure, or private-data leaks in a public issue. Use GitHub's private vulnerability-reporting form for this repository when it is available. If private reporting is unavailable, open a minimal issue asking the maintainer for a private contact channel and do not include sensitive details.

Include the affected commit, operating system, configuration, reproduction steps, impact, and any suggested mitigation when it is safe to do so. Do not include live credentials, private messages, host details, or user/project data in the report.

Security model

Cortex coordinates provider CLIs and may be configured to access local files, remote hosts, messaging systems, credentials, and write-capable agent roles. It is an operator-controlled framework, not a hostile multi-tenant isolation boundary. A role's instructions and sandbox reduce accidental scope, but they do not make elevated provider or worker access harmless.

Before enabling unattended or elevated operation:

  • use the narrowest role and permission level that can perform the work
  • keep credentials in the private secrets locations documented by Cortex
  • keep user, environment, project, inbox, log, and runtime state out of public Git history
  • configure backups and version control for important work
  • publish only through scripts/sync_public.sh, never by pushing the live operational branch
  • review third-party provider, messaging, and remote-host permissions

If you discover that private data or a credential has entered public Git history, treat the value as compromised, rotate or revoke it immediately, and report the exposure privately.

There aren't any published security advisories