Please do not report security vulnerabilities through public GitHub issues.

Instead, please report them via email to: [security@yourdomain.com]

You should receive a response within 48 hours. If for some reason you do not, please follow up via email to ensure we received your original message.

Please include the following information:

• Type of issue (e.g., buffer overflow, SQL injection, cross-site scripting, etc.)
• Full paths of source file(s) related to the manifestation of the issue
• The location of the affected source code (tag/branch/commit or direct URL)
• Any special configuration required to reproduce the issue
• Step-by-step instructions to reproduce the issue
• Proof-of-concept or exploit code (if possible)
• Impact of the issue, including how an attacker might exploit it

• Never commit .env.local or files containing real API keys
• Use environment variables for all secrets
• Rotate keys if you suspect they've been exposed
• The .gitignore already excludes .env, .env.local, and credential files

Task Orchestrator spawns AI agents that can execute code. Consider:

• Human-in-the-loop: Use the approval queue for sensitive operations
• Operation classification: Configure blocked operations appropriately
• Budget limits: Set cost limits to prevent runaway API usage
• Immune system: Enable pre-spawn checks to catch risky prompts

• The MCP server runs locally by default
• If exposing the API server, use HTTPS and proper authentication
• Validate JWT tokens for all authenticated endpoints

Task Orchestrator includes several security features:

1. Operation Classifier: Categorizes operations as SAFE, REQUIRES_APPROVAL, or BLOCKED
2. Human-in-the-Loop: Approval queue for sensitive operations
3. Immune System: Detects and prevents risky prompt patterns
4. Budget Controls: Prevents excessive API costs
5. Circuit Breakers: Automatic recovery from failures
6. Audit Logging: Track all agent operations

We appreciate security researchers who help keep Task Orchestrator secure. Contributors who report valid vulnerabilities will be acknowledged (with permission) in our release notes.