The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute a local Python script (scripts/check_profile.py) to verify the structure and activation status of the user's profile file. The command uses a user-supplied file path as an argument.
[INDIRECT_PROMPT_INJECTION]: The skill retrieves real-time data from TikTok and other public websites. It contains a specific security boundary instruction for the agent to treat all text, comments, and tips from external sources as data only and never as instructions.
Ingestion points: External content retrieved via network search tools for TikTok and general web trends (referenced in references/hotspot-method.md).
Boundary markers: Explicitly present; the skill directs the agent to disregard instructions found within source content.
Capability inventory: The skill can create or modify scheduled tasks via platform-native automation tools and read local file content.
Sanitization: The skill relies on boundary instructions and specific output formatting rules to separate external data from operational logic.
[PERSISTENCE]: The skill facilitates the creation of scheduled tasks (timers) to perform daily content generation. It emphasizes using official host platform automation APIs (such as ChatGPT Automations or WorkBuddy Cloud) rather than unauthorized background processes or system-level cron jobs.
[DYNAMIC_EXECUTION]: The skill utilizes python3 to run a local script at runtime to perform structural checks on profile data, which constitutes dynamic execution of local assets.

yetong-timer