dotnet-realtime-communication
Pass
Audited by Gen Agent Trust Hub on Mar 5, 2026
Risk Level: SAFE
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references standard, well-known libraries and services including the 'StreamJsonRpc' NuGet package and 'Azure SignalR Service'. These are official components in the .NET ecosystem used for their intended purposes.
- [DATA_EXFILTRATION]: No evidence of hardcoded credentials or unauthorized data transmission was found. The skill demonstrates best practices for handling sensitive data, such as retrieving connection strings from 'IConfiguration' and securely handling JWT tokens during WebSocket handshakes.
- [PROMPT_INJECTION]: The 'Agent Gotchas' section contains benign instructional guidance intended to improve the quality of the AI's technical advice (e.g., protocol selection and configuration warnings). No instructions attempting to bypass safety filters or override system prompts were detected.
- [COMMAND_EXECUTION]: The provided C# code snippets are focused on application logic and hub configuration. There are no patterns involving the execution of shell commands, subprocess spawning, or arbitrary code evaluation.
- [INDIRECT_PROMPT_INJECTION]: The skill defines surfaces for ingesting external data (e.g., SignalR Hub methods like 'SendMessage' and 'UploadData'). While these are standard communication patterns, the skill does not contain logic that would process this data in a way that executes embedded instructions within the agent context. No sanitization is explicitly shown in the snippets, as they are simplified architectural examples.
Audit Metadata