We take security seriously. If you discover a security vulnerability, please report it responsibly.

1. Do NOT create a public GitHub issue for security vulnerabilities
2. Email security concerns to the repository maintainers
3. Include as much detail as possible:
   • Description of the vulnerability
   • Steps to reproduce
   • Potential impact
   • Suggested fix (if any)

• Acknowledgment: We'll acknowledge receipt within 48 hours
• Assessment: We'll assess the vulnerability and its impact
• Resolution: We'll work on a fix and coordinate disclosure
• Credit: We'll credit reporters in the release notes (unless you prefer anonymity)

This tool processes user-provided paths and URLs. We implement:

• Path validation: Prevents path traversal attacks (../ sequences)
• URL validation: Only allows HTTPS GitHub URLs with strict format checking
• Command injection prevention: Uses spawnSync with array arguments instead of string interpolation

• Foundry Local runs locally on localhost:5273 by default
• No credentials are transmitted to external services
• Model responses are processed locally

• The tool reads repository contents for analysis
• Output files are written to user-specified directories
• No sensitive file access outside the target repository

We regularly audit dependencies for known vulnerabilities:

npm audit

When using this tool:

1. Review generated content before sharing publicly
2. Don't include secrets in repositories being analyzed
3. Keep dependencies updated with npm update
4. Run security audits with npm audit

Security patches will be released as soon as fixes are available. Monitor the repository releases for updates.

This security policy covers:

• The core repo-onboarding-pack tool
• The web UI server component
• Official documentation

Out of scope:

• Third-party dependencies (report to respective maintainers)
• Foundry Local itself (report to Microsoft)
• User-generated content

Before submitting code:

• No hardcoded credentials or secrets
• User inputs are validated
• File paths are sanitized
• External URLs are validated
• No use of eval() or similar
• Dependencies are from trusted sources
• Sensitive data is not logged

We appreciate responsible disclosure from security researchers helping keep this project secure.