We take the security of VT Code seriously. If you discover a security vulnerability, we appreciate your responsible disclosure and will work to address it promptly.
Please do not report security vulnerabilities through public GitHub issues.
Instead, please report security vulnerabilities via one of the following channels:
- Email: security@vtcode.org (replace with actual email if available)
- GitHub Private Vulnerability Reporting - This is the preferred method for reporting vulnerabilities, as it allows for secure, private communication.
When reporting a security vulnerability, please provide us with the following information:
- A brief description of the vulnerability and its potential impact
- Steps to reproduce the issue (POC code is appreciated)
- Affected versions (if known)
- Any possible mitigations you've identified
- Acknowledgment: We will acknowledge your report within 48 hours
- Updates: We will provide regular updates on the status of the vulnerability and fix progress
- Resolution: We will work to fix the vulnerability as quickly as possible and coordinate the release of the fix with you
- Credit: We will publicly acknowledge your responsible disclosure (unless you prefer to remain anonymous)
- Never commit API keys, tokens, or other sensitive credentials to version control
- Use environment variables for storing API keys instead of hardcoding them
- Consider using
.envfiles with proper gitignore configuration - Rotate your API keys regularly
- Keep your
vtcode.tomlconfiguration file secure and avoid sharing sensitive values - Regularly review your tool policies to ensure only necessary operations are allowed
- Use secure connections when integrating with external services
- Only run VT Code in trusted environments
- Be cautious when executing code or commands suggested by the AI agent
- Regularly update VT Code to the latest version to ensure you have the latest security patches
| Version | Supported |
|---|---|
| 0.31.x | Latest |
| 0.30.x | Critical fixes only |
| < 0.30 | No longer supported |
VT Code includes several built-in security features:
- Path Validation: Prevents file system access outside the designated workspace
- Tool Policies: Configurable allow/deny/prompt policies for different operations
- PII Protection: Automatic tokenization of sensitive data in code execution
- Token Management: Secure handling of API keys and authentication tokens
VT Code enforces the following CI/CD security controls for GitHub Actions:
- Forbidden trigger policy:
pull_request_targetandworkflow_runare blocked by policy checks. - Immutable action pinning: every third-party and first-party action in workflows is pinned to a full 40-character commit SHA.
- Workflow policy check in CI:
scripts/check_workflow_security.shruns in CI and can be run locally via./scripts/check.sh workflow-security. - Least-privilege token permissions: workflows default to
permissions: {}and grant only job-level minimum permissions. - Checkout credential hardening: all
actions/checkoutusage setspersist-credentials: falseunless explicitly required. - Release credential isolation: publishing runs in the dedicated
releasedeployment environment (with reviewer gate), rather than broad repository-level credential exposure. - No release-path cache: release workflows do not use dependency/build cache restore to reduce cache-poisoning exposure on publish paths.
For information about VT Code's security architecture, please see our documentation on:
We regularly update dependencies and monitor for security vulnerabilities in our dependencies. To check for known vulnerabilities in Rust dependencies, you can run:
# Install cargo-audit if you haven't already
cargo install cargo-audit
# Audit dependencies for known vulnerabilities
cargo auditFor general security questions or concerns, please contact us via the channels mentioned above.
Thank you for helping keep VT Code and its users safe!