The Agent Skills Directory

[SAFE]: The skill does not contain any detected prompt injections, malicious code, or unauthorized data access attempts. The instructions are aligned with the stated purpose of assisting in NVIDIA skill discovery.
[EXTERNAL_DOWNLOADS]: The skill references official catalog resources at https://build.nvidia.com/skills and https://raw.githubusercontent.com/NVIDIA/skills/main/skills.sh.json. As these originate from a trusted vendor, they are considered safe for their intended use in discovery.
[COMMAND_EXECUTION]: The skill provides instructions for the npx skills add command to list and manage skills. This is the official interface for the agent's skill management system, and the skill correctly requires explicit user confirmation before executing any installation commands.
[INDIRECT_PROMPT_INJECTION]: The skill evaluates external catalog data to provide recommendations. This attack surface is mitigated by the use of verified vendor-controlled sources and explicit instructions that prevent the agent from autonomously implementing instructions found within the data.

nvidia-skill-finder