The Agent Skills Directory

Indirect Prompt Injection Surface: The skill is designed to ingest and process review comments left within running applications, which represents a potential surface for indirect prompt injection if those comments contain adversarial instructions.
Ingestion points: The skill reads data using winapp devtools comments list and get commands in SKILL.md.
Boundary markers: There are no explicit instructions for the agent to use delimiters or ignore embedded instructions when reading comments.
Capability inventory: The skill has the capability to execute shell commands (winapp run), modify application properties (set-property), and suggests editing local source files.
Sanitization: No sanitization or validation of the comment content is mentioned before the agent acts upon it.
Local Command Execution: The skill relies on executing local commands via the winapp CLI to manage application lifecycles and perform inspections.
Use of winapp run and winapp devtools allows the agent to interact directly with the local development environment and running processes.
External Resource References: The skill references external documentation from the official Microsoft GitHub repository. These references are used to provide the user with additional guidance on advanced developer features.

winui-devtools