The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/fetch_xhs_ai.py utilizes the subprocess module to execute system commands. Specifically, it uses subprocess.run to call launchctl on macOS and crontab on Linux for managing automated tasks. It also executes open or xdg-open to launch the generated HTML report in the user's browser.
[PERSISTENCE]: The skill implements a subscription feature in scripts/fetch_xhs_ai.py that establishes persistence by creating a macOS LaunchAgent at ~/Library/LaunchAgents/com.qoder.xiaohongshu-ai-feed.plist or adding a new entry to the user's crontab. These tasks are configured to run the skill's main script daily at 16:00.
[INDIRECT_PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface.
Ingestion points: Data is fetched from an external API (https://redfox.hk/story/api/parseWork/queryXhsAiMsgs) containing Xiaohongshu post titles, usernames, and content tags.
Boundary markers: None. The fetched data is directly interpolated into an HTML template.
Capability inventory: The skill has the ability to write files to the local file system (~/Downloads/QoderReports/) and execute shell commands via subprocess.
Sanitization: There is no evidence of HTML escaping or sanitization of the data retrieved from the API before it is rendered into the 小红书AI日报_*.html report, which could allow for script injection (XSS) if the source data contains malicious payloads.

xiaohongshu-ai-feed