The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill exhibits an indirect prompt injection risk surface:
Ingestion points: The agent reads untrusted content from the project workspace, specifically checking the .csproj file to extract the TargetFramework element.
Boundary markers: Absent. There are no delimiters or instructions to ignore potential embedded injection markers within the parsed project files.
Capability inventory: The skill writes local files (Dockerfile, .dockerignore, progress.md) and executes local terminal commands via shell invocation (docker build).
Sanitization: Absent. No input validation or escaping mechanisms are specified for values parsed from the workspace before being interpolated into files or shell execution commands.
[COMMAND_EXECUTION]: The skill instructs the agent to execute a shell command (docker build -t aspnetcore-app:latest .) to verify containerization success. If project files or the generated Dockerfile are subverted, arbitrary commands could execute during the Docker build sequence.
[EXTERNAL_DOWNLOADS]: The skill references links to official documentation and release information from the Microsoft .NET Docker repository (github.com/dotnet/dotnet-docker). These references are used solely for identifying valid image tags and do not pose a threat.

containerize-aspnetcore