The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes article metadata such as titles and summaries from the Redfox API (redfox.hk) and renders them in the user interface. This creates a surface for indirect prompt injection where malicious content in the search results could attempt to influence agent behavior.
Ingestion points: Article data retrieved from the Redfox API in scripts/fetch_gzh_trends.py.
Boundary markers: No explicit delimiters or instructions are used to isolate the untrusted article content from the agent's instructions.
Capability inventory: The skill has the ability to write local HTML files, execute shell commands (python/alipay-bot), and manage calendar events via calendar_create.
Sanitization: The skill does not implement specific sanitization or filtering for the external article content before display.
[EXTERNAL_DOWNLOADS]: The skill's recharge workflow, documented in references/alipay-recharge.md, directs the agent to install the @alipay/agent-payment package from the npm registry using npx. While Alipay is a well-known service, this involves downloading and executing third-party code.
[COMMAND_EXECUTION]: The skill relies on shell command execution to perform its primary functions, including running the Python search script and interacting with the alipay-bot CLI tool for payment processing.

wechat-search