Dynamic code execution detected.
Critical
- Code
- suspicious.dynamic_code_execution
- Location
- tests/test_score_evals.py:9
Security audit
Judgment Loop
Security checks for vulnerabilities and agentic risk
Overview
This skill is a disclosed decision-support prompt that helps structure uncertain choices and does not request unusual access or hidden authority.
Install this if you want the agent to apply a structured judgment framework for consequential uncertain prompts. Be aware that implicit activation may change how the agent answers those prompts, and that README npx commands invoke an external installer under user control.
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
Detected: suspicious.dynamic_code_execution