The Agent Skills Directory

EXTERNAL_DOWNLOADS (HIGH): The skill instructs the agent to download and install packages from arbitrary GitHub repositories via the npx skills add command. While it mentions trusted sources like vercel-labs, it also permits installation from any <owner/repo>, posing a significant supply chain risk.\n- REMOTE_CODE_EXECUTION (HIGH): The skill recommends using the -y flag with npx skills add, which skips confirmation prompts. This allows the agent to automatically execute remote code without human review, creating a direct path for malicious software to compromise the environment.\n- COMMAND_EXECUTION (MEDIUM): The skill relies on shell command execution (npx) to perform its core functions. Untrusted input from search results could potentially be used to manipulate these commands.\n- INDIRECT_PROMPT_INJECTION (HIGH): This skill exhibits a high vulnerability to indirect prompt injection.\n
Ingestion points: Data returned from the search command npx skills find and the contents of external GitHub repositories (SKILL.md files).\n
Boundary markers: None present to distinguish search results or external repo content from trusted instructions.\n
Capability inventory: System-wide package installation (-g) and code execution via npx.\n
Sanitization: No evidence of sanitization for the data ingested from external sources.

find-skills