The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill's Dockerfile templates (assets/template-Dockerfile.amd64 and assets/template-Dockerfile.amd64.debug) contain instructions to download the Microsoft artifact credential provider and the VS debugger. These downloads target official and well-known Microsoft resources (raw.githubusercontent.com/Microsoft/ and aka.ms).
[COMMAND_EXECUTION]: The skill executes local Python scripts and shell commands to perform project structure detection, manifest updates, and solution file management. These operations are performed on the local filesystem using inputs derived from the project's own files or direct user prompts. Shell arguments are properly quoted to mitigate common injection risks.
[INDIRECT_PROMPT_INJECTION]: The skill acts as an automated scaffolding tool that ingests data from existing project files (such as C# source code, Dockerfiles, and deployment manifests) to determine configuration values.
Ingestion points: Files like Program.cs, module.json, and *.deployment.manifest.json are scanned by scripts/detect_project_structure.py and scripts/scan_manifests.py.
Boundary markers: The skill does not use explicit delimiters when interpolating detected project data into new files, but it relies on strict templates in the assets/ directory.
Capability inventory: Capabilities include file creation, directory creation, and modification of existing manifest and solution files via subprocess calls.
Sanitization: The skill relies on the user to confirm detected project settings (Step 1.5) and normalized module names (Step 3) before proceeding with file generation.

azure-iot-edge-module