The Agent Skills Directory

[SAFE]: The skill implements robust operational safeguards, including a hard cap of 10 iterations and mandatory progress tracking, which prevents the agent from entering infinite retry loops or consuming excessive resources.
[COMMAND_EXECUTION]: The skill utilizes standard, local .NET development commands (dotnet build, dotnet test) as part of its core functionality to verify code integrity.
[PROMPT_INJECTION]: No malicious instruction overrides or safety filter bypasses were identified. The instructions use standard natural language to guide agent logic for debugging and refactoring.
[DATA_EXFILTRATION]: There is no evidence of network activity or commands designed to access and exfiltrate sensitive data such as credentials or environment variables.

autonomous-loops