The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses Bash(grep) and Bash(perl) to identify and remove zero-width characters, soft hyphens, and other invisible Unicode sequences from files. These commands are scoped to specific patterns and are a functional part of the 'humanizing' process.
[DATA_EXPOSURE_PREVENTION]: The skill instructs the agent to read video-agent/voice.md if it exists. This follows standard practices for local configuration and voice preferences without attempting to access sensitive system files (like SSH keys or AWS credentials).
[NO_CODE_INJECTION]: While the skill uses regular expressions to filter Unicode tags and joiners, it does not execute dynamically generated code from external or untrusted sources.
[PROMPT_INJECTION_SAFE]: The instructions provide guidance on how to mimic a human creator's voice. It does not attempt to override the underlying AI safety filters or bypass system personas, but rather guides the tone and vocabulary of the output.

video-human