The Agent Skills Directory

[UNVERIFIABLE_DEPENDENCIES_AND_REMOTE_CODE_EXECUTION]: The skill instructs the user to install a command-line interface via npx, which downloads and executes code from a remote registry.
Evidence: npx -y @mvanhorn/printing-press install openart --cli-only in SKILL.md.
Source: The package is associated with the author's namespace (@mvanhorn).
[DATA_EXPOSURE_AND_EXFILTRATION]: The skill requires the use of tools like pycookiecheat or cookies to extract browser session data (cookies) for authentication, which involves accessing sensitive application data.
Evidence: openart-pp-cli auth login --chrome requiring pycookiecheat via pip or cookies via Homebrew in SKILL.md.
[DATA_EXFILTRATION]: The skill provides a mechanism to deliver command output directly to external URLs via webhooks.
Evidence: --deliver webhook:<url> flag described in the Output Delivery section of SKILL.md.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data that could contain malicious instructions, such as generation history and user-provided prompts for enhancement.
Ingestion points: Data enters the context via user prompts, the history command, and the media list command.
Boundary markers: None specified to distinguish between data and instructions.
Capability inventory: The skill can execute shell commands (openart-pp-cli), write files (--download), and perform network operations (webhook delivery).
Sanitization: No explicit sanitization or filtering of external content is mentioned.

pp-openart