If you discover a security vulnerability in this plugin, please report it responsibly:

1. Do not open a public issue
2. Email the maintainers directly or use GitHub's private vulnerability reporting feature
3. Include a detailed description of the vulnerability
4. Provide steps to reproduce if possible

We will respond within 48 hours and work with you to understand and address the issue.

This plugin is designed with security in mind:

• No external dependencies — All functionality is self-contained within Claude Code
• No data collection — Your prompts and outputs stay local to your Claude Code session
• No network calls — The plugin itself makes no network requests; all API calls are handled by Claude Code
• Open source — Full transparency for security audits

The Argus persona helps design defenses against prompt injection attacks in your agentic systems:

• Input validation and sanitization
• Structured output enforcement
• Capability restrictions
• Trust boundary definitions

This plugin:

• Does not store any user data
• Does not log prompts or outputs
• Does not transmit data to external services
• Operates entirely within Claude Code's security model

1. Review generated architectures — Always review security recommendations before implementation
2. Validate outputs — Use Argus outputs as a starting point, not a final solution
3. Keep Claude Code updated — Security improvements are delivered through Claude Code updates
4. Report issues — Help us improve by reporting any security concerns

Specifically designed for security architecture:

• Threat modeling
• Zero trust architecture
• Defense in depth patterns
• Authentication/authorization design
• Audit logging specifications
• Prompt injection defenses

Every persona includes:

• Safety guardrails in system prompts
• Scope restrictions to prevent off-topic behavior
• Clear boundaries for agent capabilities