The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external information from a 'project marketing context file', which creates a surface for potential indirect prompt injection attacks.\n- Ingestion points: In SKILL.md, the agent is instructed to read a 'project marketing context file' before starting the task.\n- Boundary markers: No specific delimiters or instructions are provided to separate the ingested context from the agent's primary instructions.\n- Capability inventory: The skill focus is limited to generating marketing copy; no dangerous capabilities like shell command execution or network access are requested.\n- Sanitization: The instructions do not define any validation or filtering mechanisms for the data found in the external context file.\n- [SAFE]: The skill consists of standard marketing and writing principles and does not request sensitive permissions or perform any malicious operations.

copywriting