The Agent Skills Directory

[DATA_EXFILTRATION]: The skill instructions direct the agent to access user-specific context files at .liftli/voice.md and ~/.liftli/voice.md to personalize the generated content. These files are used to store user voice and style preferences.
[COMMAND_EXECUTION]: The skill provides instructions for users to manually run commands such as npx skills add liftli-ai/linkedin-agent-skills and claude mcp add to install additional vendor tools. These commands target resources within the author's ecosystem.
[EXTERNAL_DOWNLOADS]: The skill references external resources hosted on liftli.ai and mcp.liftli.ai for tool configuration and documentation. These resources are provided by the verified author of the skill.
[INDIRECT_PROMPT_INJECTION]: The skill acts as a surface for indirect prompt injection because it ingests untrusted data (LinkedIn posts) provided by the user to generate replies. However, the skill lacks autonomous execution capabilities that would allow such injections to perform sensitive operations.

linkedin-comment-generator