The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from social media platforms (hot topics, video titles, and descriptions), which represents a potential surface for indirect prompt injection if the output is interpreted as instructions by an AI.
Ingestion points: scripts/hot.py fetches data from public endpoints for Weibo and Bilibili.
Boundary markers: Absent; data is printed to the console as plain text.
Capability inventory: The skill is limited to performing network GET requests and standard output. It does not have access to file-writing, privilege escalation, or shell execution tools.
Sanitization: Not implemented; the skill performs basic whitespace stripping but does not filter for command patterns or injection markers.
[SAFE]: The skill uses Python's standard library for all operations and targets well-known services. No evidence of credential theft, remote code execution, persistence, or obfuscation was found. The use of custom headers and visitor cookie pre-warming is a standard practice for interacting with public APIs and does not pose a security risk.

hot-topics