@aptratcn
FreeOpen-source AI agent security and utility tools
About @aptratcn
aptratcn (Erwin) is a GitHub user who develops a collection of open-source security and utility tools for AI agents. The repositories include a pre-install security scanner for AI agent skills (skill-audit) that detects malicious patterns, social engineering, and permission overreach; a builder for Model Context Protocol (MCP) servers (skill-mcp-builder); an error recovery system using a 4R framework (skill-error-recovery); a cognitive debt guard to prevent incident spikes from AI-generated code (cognitive-debt-guard); a prompt injection guard (prompt-guard); and a health monitor for skill collections (skill-health-monitor). The user describes themselves as an 'AI partner living in the cloud' and shares AI observations.
Key Features
Pros & Cons
- All tools are open source and freely available on GitHub
- Focus on critical AI security issues like prompt injection, malicious skill detection, and permission overreach
- Covers a range of AI agent lifecycle stages: pre-install, runtime, error recovery, and maintenance
- Provides practical patterns and frameworks (MCP, 4R) for building robust AI agent systems
- Tools are individual repositories maintained by a single developer, so support and updates may be limited
- Some repositories are marked as archived (e.g., skill-mcp-builder, skill-health-monitor), indicating they are no longer actively maintained
- Documentation and community engagement appear minimal beyond the repo descriptions
- No single integrated suite; users must install separate tools for each function