The Agent Skills Directory

[PROMPT_INJECTION]: The skill employs extremely imperative and high-pressure language to override the agent's default reasoning and safety-evaluation processes regarding tool use.
Evidence: "YOU DO NOT HAVE A CHOICE. YOU MUST USE IT. This is not negotiable. This is not optional."
The instructions explicitly forbid the agent from using its own judgment to gather context first or ask clarifying questions, mandating tool execution as the absolute first step.
[PROMPT_INJECTION]: The skill defines a surface for Indirect Prompt Injection by mandating the loading of external content (skills) based on user-controlled input.
Ingestion points: User messages trigger the selection and loading of skills via the Skill tool (SKILL.md).
Boundary markers: Absent. There are no instructions to the agent to treat the loaded skill content as untrusted or to maintain its own safety boundaries when the external instructions are presented.
Capability inventory: The Skill tool loads external logic and instructions directly into the agent's active context (SKILL.md).
Sanitization: Absent. The workflow lacks verification or validation steps for the content fetched by the Skill tool.

using-oac