The Agent Skills Directory

System Tool Interaction: The skill instructs the agent to execute various build and management tools, including dotnet, xcrun simctl, and maui-devflow. These operations are essential for the skill's purpose but involve the execution of shell commands that interact with the host system.
Configuration and Secret Access: The protocol directs the agent to locate and use application secrets or API keys by inspecting files such as appsettings.json, .env files, and environment variables. This is intended for local app verification but represents a pattern of accessing sensitive configuration data.
High-Autonomy Directives: The instructions use directive language ("Rule 1: Own Your Dependencies", "fix it", "never skip") to discourage the agent from reporting typical errors or blockers. Instead, it is instructed to autonomously modify its environment, such as building tools from source or creating new simulator instances, which increases the complexity of the agent's actions.
Indirect Prompt Injection Surface: The skill requires the agent to ingest and analyze external data, including project source code at ~/work/BaristaNotes and screenshots at ~/Downloads/. This interaction with external, potentially untrusted data represents an attack surface for indirect instructions, although no specific malicious content is present in the skill itself.

verification-protocol