The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a provided Python script (speaker_names.py) to perform its analysis. The script uses standard libraries and implements safe file handling practices, such as using temporary files for atomic writes and refusing to overwrite existing files.
[DATA_EXPOSURE_AND_EXFILTRATION]: The skill instructions provide clear guidance on handling sensitive metadata, explicitly advising against persisting private meeting links or access tokens. The script itself performs no network operations and only interacts with local files provided as command-line arguments.
[INDIRECT_PROMPT_INJECTION]: While the skill processes potentially untrusted transcript data, it implements a multi-step workflow (inspect then apply) that requires explicit user-defined mappings. This structure ensures that findings are presented to the agent for verification rather than automatically executed, mitigating risks associated with malicious text in the input data.

edgespeak-name-speakers