Skip to content

Security: GabrielHenriqueCA/cortex-ai

Security

SECURITY.md

Security Policy

Supported Versions

Version Supported
0.3.x (latest)
0.2.x ✅ security fixes only
0.1.x

Reporting a Vulnerability

Do not open a public GitHub issue for security vulnerabilities.

Report vulnerabilities privately via email: security@gabrielh.dev

Include in your report:

  • Description of the vulnerability
  • Steps to reproduce
  • Potential impact (what an attacker could achieve)
  • Any suggested remediation

What to expect

Timeframe Action
48 hours Initial acknowledgment of your report
7 days Preliminary assessment and severity classification
30 days Status update on remediation progress
90 days Coordinated disclosure deadline

We follow a 90-day coordinated disclosure timeline. After 90 days, we may publish details of the vulnerability regardless of whether a fix has been released, to protect the community.

Scope

This template contains:

  • Shell scripts (.claude/hooks/*.sh, scripts/*.sh)
  • Configuration files (.pre-commit-config.yaml, .claude/settings.json)
  • Documentation and prompt files

In scope:

  • Hooks that execute arbitrary commands with unintended side effects
  • Configuration that grants excessive permissions
  • Prompt injection vulnerabilities in agent definitions
  • Scripts that handle sensitive data insecurely

Out of scope:

  • Issues in third-party tools (Graphify, pre-commit, ruff, prettier) — report to those projects directly
  • Social engineering attacks
  • Theoretical vulnerabilities without a concrete exploit path

Security Best Practices for Users

  • Review .claude/settings.json permissions before using in a sensitive codebase
  • Use claude --sandbox mode for untrusted projects
  • Never commit .env files or credentials — use CLAUDE.local.md and .claude/settings.local.json (both gitignored) for local secrets
  • Run gitleaks detect before pushing if you've added credentials during a session

There aren't any published security advisories