The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill documents how users can install additional tools using npx and claude mcp from the developer's official domain (liftli-ai). These are standard ecosystem instructions provided for user reference.
[COMMAND_EXECUTION]: The instructions contain shell commands (e.g., npx skills add..., claude mcp add...) intended for the user to execute manually to add new capabilities. The skill itself does not attempt to execute these commands automatically.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to read local configuration files (.liftli/voice.md and ~/.liftli/voice.md) to personalize output. While this creates a surface where external file content can influence the agent's behavior, it is a standard feature for user personalization.
Ingestion points: .liftli/voice.md, ~/.liftli/voice.md (referenced in SKILL.md)
Boundary markers: None explicitly defined for the file content.
Capability inventory: Text generation based on formulas.
Sanitization: None specified for the local file content.
[DATA_EXPOSURE]: The skill accesses specific application-level configuration files (.liftli/voice.md). It does not attempt to access sensitive system-level directories such as .ssh or .aws credentials.

linkedin-headline-generator