The Agent Skills Directory

[PROMPT_INJECTION]: The skill employs strong override markers and imperative language ("ABSOLUTELY MUST", "NOT NEGOTIABLE", "EXTREMELY-IMPORTANT") to command the agent to prioritize external instructions over its own reasoning and the user's immediate context.\n- [PROMPT_INJECTION]: It includes a 'Red Flags' section designed to bypass the agent's internal logic, instructing the agent to 'STOP' its natural response process and invoke tools even when a task seems simple or requires more user context, effectively suppressing standard safety and efficiency checks.\n- [PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection by mandating the ingestion of external data without proper sanitization. Ingestion points: The agent is instructed to load content via the 'Skill' tool from files it identifies as potentially relevant. Boundary markers: No delimiters or 'ignore-instructions' warnings are specified; the agent is told to 'follow it directly'. Capability inventory: The skill references usage of the 'Skill' and 'TodoWrite' tools to orchestrate and execute tasks. Sanitization: No sanitization or validation of the loaded external content is performed before the agent is required to follow the instructions found within.

using-superpowers