The Agent Skills Directory

[PROMPT_INJECTION]: The skill is designed to read and respect instructions from a local AGENTS.md file, which creates a surface for indirect prompt injection if that file contains malicious directives meant to influence the agent's behavior or command selection.
Ingestion points: Reads AGENTS.md (documented in SKILL.md under compatibility and quick start).
Boundary markers: None explicitly defined to separate instructions in AGENTS.md from the skill's own logic.
Capability inventory: The skill performs package installations (dotnet add package), tool installations (dotnet tool install), and command execution (dotnet test, dotnet tool run).
Sanitization: No explicit sanitization or validation of the content found within AGENTS.md is described.

dotnet-coverlet