The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructs the agent to run commands that download and execute code from an external source using npx lovstudio skills add proposal and uvx lovstudio-skill-helper decrypt proposal. These tools originate from the skill author's infrastructure but are not part of a globally trusted registry.
[OBFUSCATION]: The core logic and instructions of the skill are stored in an encrypted file SKILL.md.enc using the AES-256-GCM cipher. This prevents the analysis of the actual instructions the agent will follow at runtime, posing a risk as the decrypted behavior cannot be verified beforehand.
[COMMAND_EXECUTION]: The installation and initialization process relies on direct shell command execution to manage the lifecycle of the skill, including decryption and credential management.

lov-proposal