The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/new_deck.py utilizes subprocess.run to open generated .iapresenter bundles in the application. This is a standard macOS automation pattern using a fixed binary path and does not involve shell execution.
[COMMAND_EXECUTION]: The script scripts/preview_deck.sh uses the macOS screencapture utility to allow an agent to verify rendered slide output. This functionality requires explicit system-level Screen Recording permissions and is restricted to its documented purpose.
[DATA_EXPOSURE]: The skill interacts with the user's Library folder (~/Library/Containers/net.ia.presenter/...) to manage themes. These file operations are limited to the application's specific support container and are necessary for the skill's theme installation functionality.
[INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided Markdown data through scripts/lint_deck.py and scripts/new_deck.py. Analysis shows the data is parsed for structural syntax only and is not executed or interpreted as instructions that could influence agent behavior.
[SAFE]: No evidence of prompt injection, obfuscation, hardcoded credentials, or unauthorized remote code execution was found across the skill's instructions and supporting scripts.

ia-presenter