The Agent Skills Directory

External Downloads (HIGH): The skill uses npx to download and execute taze and @antfu/ni from the npm registry. Since the author/organization antfu-collective is not on the trusted sources list, this is classified as a high-risk remote download and execution pattern.
Remote Code Execution (HIGH): The skill executes npx taze --install and npx @antfu/ni. These commands trigger package manager installation processes which execute arbitrary lifecycle scripts (e.g., preinstall, postinstall) defined in package.json.
Indirect Prompt Injection (HIGH): This skill exhibits a significant vulnerability surface. 1. Ingestion points: Reads project package.json files using cat and find in SKILL.md. 2. Boundary markers: Absent; there are no delimiters or instructions to ignore embedded commands in the files being read. 3. Capability inventory: Possesses file-write capabilities (--write) and arbitrary command execution through package installation (--install). 4. Sanitization: None; the skill does not validate or sanitize the contents of package.json before processing them. An attacker who can influence a project's package.json could execute malicious code when the agent runs this skill.

up-deps