jacobson-network-performance

Pass

Audited by Gen Agent Trust Hub on Mar 9, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection through its diagnostic utility implementations.
  • Ingestion points: The target parameter in the Traceroute class and the samples list in the NetworkDiagnostics class are entry points for potentially untrusted data.
  • Boundary markers: There are no boundary markers, delimiters, or explicit instructions to ignore embedded commands within the processed data.
  • Capability inventory: The skill includes code to perform low-level network operations using socket and socket.SOCK_RAW across multiple classes.
  • Sanitization: No input validation or sanitization is performed on the network targets or the diagnostic samples before processing.
  • [COMMAND_EXECUTION]: The provided Traceroute implementation utilizes socket.SOCK_RAW. This is a privileged operation that typically requires root or Administrator permissions to execute. In an automated environment, attempting to run this code without appropriate permissions would result in system errors, and it represents a capability to perform high-privilege network interactions.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 9, 2026, 10:49 PM