The Agent Skills Directory

[DATA_EXPOSURE_AND_EXFILTRATION]: The skill manages authentication for the target scraping site and the Gemini API via a local .env file. This is a standard and recommended practice for protecting credentials in local development environments.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted content from the jiuyangongshe.com website. This content is processed by an LLM to extract structured data. The skill mitigates risks by using yaml.safe_load() to parse the AI output and limiting the use of extracted data to database storage and natural language reporting, preventing it from influencing sensitive system operations.
[EXTERNAL_DOWNLOADS]: The skill fetches data from the intended target domain (jiuyangongshe.com) and utilizes the official Google Gemini API for data processing. These are legitimate operations consistent with the skill's stated purpose.

jiucai-capture