The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external data (Xiaohongshu note titles and descriptions) fetched from the Redfox API. This data is then rendered by the agent, creating a surface for indirect prompt injection if the retrieved content contains malicious instructions aimed at the agent.
Ingestion points: Data is ingested through the scripts/fetch_xhs_hot_articles.py script, which retrieves JSON data from the redfox.hk API endpoint.
Boundary markers: The SKILL.md instructions enforce strict Markdown table formatting to delimit search results.
Capability inventory: The skill executes a local Python script with network and file-writing permissions and can invoke the calendar_create tool based on user input.
Sanitization: The Python script includes basic HTML entity escaping (converting < to <, etc.) when generating the visualization report file, providing some protection for the output.
[COMMAND_EXECUTION]: The agent is instructed to execute the local Python script scripts/fetch_xhs_hot_articles.py using the system shell to perform data searches and generate reports.
[EXTERNAL_DOWNLOADS]: The skill's core script makes network requests to the vendor's official API domain (redfox.hk) to fetch trending data. This is a standard functional requirement for the skill's operation.

xiaohongshu-search