Security audit
Claude Code Patterns
Security checks for vulnerabilities and agentic risk
Overview
This is a documentation-only skill about AI agent engineering patterns and does not install code or request access to private data.
Install this as reference material. If you copy its patterns into production, separately review any tool execution, keychain or MDM access, speculative execution, context deletion or summarization, and retry logic for explicit user control and data minimization.
Vulnerability Patterns
- Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
- Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
- Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
- Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
- Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Static analysis
No suspicious patterns detected.