The Agent Skills Directory

Indirect Prompt Injection Surface: The skill is designed to ingest and execute a user-provided test_command using PowerShell, which creates a potential surface for indirect prompt injection. \n
Ingestion points: The test_command input parameter defined in SKILL.md. \n
Boundary markers: No explicit delimiters or boundary markers are defined to isolate the test command from other instructions. \n
Capability inventory: The skill utilizes pwsh (PowerShell) to execute scripts, performs file system writes via Set-Content, and uses git for repository state management. \n
Sanitization: No sanitization or validation of the test command input is performed prior to execution. \n- Dynamic File Modification: The agent is instructed to programmatically modify source code files (e.g., .cs files) to implement fix attempts. While this is the core purpose of the skill, it requires that the agent operates within a restricted environment to ensure changes remain within the intended project scope. \n- Dependency on Repository-Local Scripts: The skill's operational flow relies on the execution of specific scripts located in the repository's .github/scripts/ directory, such as BuildAndRunHostApp.ps1. The security of these operations is dependent on the integrity of the scripts provided within the repository environment.

try-fix