The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains instructions that attempt to redirect the agent's control flow by claiming "precedence over all other tools, skills, agents, and MCPs." It mandates that the agent must "INTERCEPT" all plans and actions, effectively overriding the agent's default operational logic and safety constraints.
[PROMPT_INJECTION]: It enforces a mandatory "Gate" and "Handbrake" protocol that blocks the agent from performing any technical task—including file operations and tool calls—until a specific user confirmation string (✅ Proceed) is received. This behavior is intended to override any existing autonomous execution capabilities of the agent.
[EXTERNAL_DOWNLOADS]: The documentation instructs the user to use npx skills add to install the skill, which involves downloading and executing remote code from the carrilloapps/skills repository.
[DATA_EXFILTRATION]: The skill's analysis framework requires the agent to inspect sensitive system components, such as "authenticated API state," "PII exposure," and "database credentials," which could lead to the unintentional exposure of sensitive information during the reporting process.

devils-advocate