The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use shell commands like ls or the Glob tool to check for existing documentation in the doc/ directory. This is used solely for project state discovery and user confirmation before overwriting files.
[PROMPT_INJECTION]: The skill processes untrusted external data (QML source code, C++ headers, and build files), creating a surface for indirect prompt injection if those files contain malicious instructions.
Ingestion points: Source files including .qml, .h, CMakeLists.txt, and qmldir are read as primary inputs in SKILL.md and platform variants.
Boundary markers: The platform-specific variants (platforms/copilot.prompt.md, platforms/windsurf.md) contain explicit instructions to treat source content as technical material and ignore any instructions found within them. The main SKILL.md relies on the task description but lacks explicit markers.
Capability inventory: The skill possesses file-read access, file-write access (to the doc/ subdirectory), and limited shell command execution for file listing.
Sanitization: No explicit sanitization or escaping of the source content is mentioned before it is processed by the agent.

qt-qml-docs