The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute the feedgrab CLI tool with various parameters to fetch content from platforms like X (Twitter), YouTube, and WeChat. This is the primary intended function of the skill.
[CREDENTIALS_UNSAFE]: While the skill mentions the use of API keys (e.g., YOUTUBE_API_KEY, TWITTERAPI_IO_KEY) and session cookies, it correctly instructs the user to configure these via environment variables or a login command (feedgrab login). No hardcoded credentials or secrets are present in the skill file.
[EXTERNAL_DOWNLOADS]: The skill mentions a dependency on an external tool named feedgrab. It provides instructions for checking its presence (which feedgrab) and refers to a setup process (/feedgrab-setup), but does not contain direct download or execution of untrusted scripts from the internet.
[INDIRECT_PROMPT_INJECTION]: The skill's primary purpose is to fetch bulk content from external, user-generated sources (social media posts, search results, etc.). This inherently creates an indirect prompt injection surface if the fetched content contains instructions that the agent might inadvertently follow.
Ingestion points: Data fetched from X, YouTube, Xiaohongshu, WeChat, and Zhihu.
Boundary markers: None specified in the provided markdown file.
Capability inventory: Shell command execution via the feedgrab utility.
Sanitization: Not explicitly handled within the skill's instruction set.

feedgrab-batch