The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local Python script (search_douyin.py) to interact with its search API. It also provides instructions for setting up persistence via crontab to enable daily automated content pushes, which involves recurring shell command execution on the host system.
[DATA_EXFILTRATION]: The skill transmits user search keywords and a sensitive environment variable (REDFOX_API_KEY) to the external endpoint https://redfox.hk/story/api/dy/search/search. While this is expected vendor functionality, it involves the outbound transfer of user-derived data.
[PROMPT_INJECTION]: The skill processes untrusted external data retrieved from the Douyin API, creating a surface for indirect prompt injection.
Ingestion points: Video titles and author metadata from the search_douyin.py output are processed in the result display step.
Boundary markers: No explicit markers are used to isolate untrusted data from the agent's instructions.
Capability inventory: The agent has shell access to run local scripts and schedule persistent cron jobs.
Sanitization: There is no indication that retrieved content is sanitized or escaped before being displayed to the user or processed by the agent.

douyin-search