The Agent Skills Directory

[DATA_EXFILTRATION]: The scripts scripts/gzh_growth_fetcher.py and scripts/gen_gzh_html.py attempt to read sensitive shell configuration files, including ~/.zshrc, ~/.bashrc, ~/.bash_profile, ~/.profile, and Windows PowerShell profiles. This access is used to search for the REDFOX_API_KEY, but it exposes all contents of these configuration files to the script's memory.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection. 1. Ingestion points: External data is fetched from the RedFoxHub API via the requests library in the provided Python scripts. 2. Boundary markers: The skill lacks explicit delimiters or instructions to ignore potential commands within the fetched data when presenting it to the agent. 3. Capability inventory: The skill can perform file system writes and execute Python scripts that make network requests. 4. Sanitization: There is no evidence of sanitization for account names or other textual data returned by the API before it is processed by the agent.

wechat-top-account