protocol-analysis

Pass

Audited by Gen Agent Trust Hub on Mar 24, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes untrusted data from external network sources during protocol analysis.\n
  • Ingestion points: Network responses from APIs, WebSockets, and gRPC services analyzed via js-reverse-mcp, android_proxy_mcp, and Python requests calls (SKILL.md).\n
  • Boundary markers: No explicit delimiters or safety warnings are used to differentiate untrusted network data from the agent's core instructions.\n
  • Capability inventory: The skill utilizes shell execution (pip), Python code execution (requests, protobuf), and browser control tools (js-reverse-mcp).\n
  • Sanitization: There are no instructions for sanitizing or validating the captured network data before processing and analysis.\n- [EXTERNAL_DOWNLOADS]: The instructions suggest installing third-party Python packages including protobuf, blackboxprotobuf, and pbtk via pip. These are standard tools for security analysis but constitute external code dependencies.\n- [COMMAND_EXECUTION]: The skill provides Python code templates for replaying API requests, managing gRPC calls, and parsing binary data. These scripts are intended to be executed to verify analysis results.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 24, 2026, 04:31 AM
Security Audit — agent-trust-hub — protocol-analysis