The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user data which represents a potential surface for indirect prompt injection attacks.
Ingestion points: The inputs product_description, target_audience, and key_talking_points in SKILL.md are directly incorporated into the generation process.
Boundary markers: Absent. The skill does not use specific delimiters or instructions to the model to ignore potential commands embedded in the user inputs.
Capability inventory: The skill is limited to text generation. No file system writes, network requests, or shell command executions are present in the provided file.
Sanitization: No sanitization or validation of the user-provided input strings is performed.
[DATA_EXFILTRATION]: The skill contains a hardcoded link to an external domain for attribution.
Evidence: https://copyaccelerator.com/join in SKILL.md.
Context: This is a legitimate marketing link for the RMBC framework author and does not involve the transmission of sensitive data.

ugc-brief