The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from several markdown files to personalize content recommendations, creating a potential attack surface for indirect prompt injection.
Ingestion points: Data is read from files named identity.md, audience.md, style.md, platforms.md, and tone.md as specified in the 'Profile 感知' section of SKILL.md.
Boundary markers: There are no explicit delimiters or instructions provided to the agent to treat the content of these files as potentially adversarial data.
Capability inventory: The skill does not utilize any high-risk tools or capabilities; it has no network access, file write permissions, or shell execution requirements.
Sanitization: The skill lacks logic to validate, sanitize, or escape the content of the profile files before interpolating them into the analysis workflow.

skill-trend-rider