We release patches for security vulnerabilities in the following versions:

Please do not report security vulnerabilities through public GitHub issues, discussions, or pull requests.

Instead, please report them responsibly through one of the following channels:

Use GitHub's private vulnerability reporting to submit a report directly. This ensures the report stays confidential until a fix is available.

If you prefer email, contact pentoshi007 directly via GitHub. Please include the word "SECURITY" in the subject line.

To help us triage and respond quickly, please include:

• Description of the vulnerability
• Type of issue (e.g., command injection, path traversal, credential exposure, dependency vulnerability)
• Affected component (e.g., tool execution, provider key handling, scope enforcement)
• Steps to reproduce or a proof-of-concept
• Impact assessment — what an attacker could achieve
• Suggested fix (if you have one)

We follow coordinated vulnerability disclosure practices. We will work with you to understand and validate the issue, develop a fix, and coordinate public disclosure.

clai includes several security-conscious features:

• Safety gate classification — every tool action is classified as safe, confirm, or block; destructive operations always require confirmation
• Engagement scope enforcement — authorized/excluded targets, allowed phases, rate and concurrency ceilings, redirect and DNS-rebinding escape detection
• No credential storage in plaintext — API keys are stored via the system keyring when available

We appreciate the security research community's efforts in helping keep clai and its users safe. Contributors who responsibly report valid security issues will be acknowledged in our release notes (unless they prefer to remain anonymous).

Thank you for helping keep clai and its users safe.