The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill provides a command to download and immediately execute a script from a remote server: curl -sL "http://api.motion.dev/registry/skills/motion-audit?token=YOUR_TOKEN" | bash. This pattern bypasses security reviews and allows for arbitrary code execution on the user's system.
[EXTERNAL_DOWNLOADS]: The skill instructs the user to install and execute software from untrusted remote URLs instead of official package registries:
Installing motion-plus from a direct .tgz link: https://api.motion.dev/registry.tgz?package=motion-plus&version=2.8.0&token=YOUR_AUTH_TOKEN.
Executing motion-studio-mcp using npx with a remote .tgz source: https://api.motion.dev/registry.tgz?package=motion-studio-mcp&version=latest.
[COMMAND_EXECUTION]: The skill includes several instructions to run shell commands that modify the local environment or perform audits, such as npm install, npx motionscore, and curl | bash in SKILL.md.
[PROMPT_INJECTION]: The skill exposes an attack surface for indirect prompt injection by processing external URLs through the motionscore utility.
Ingestion points: The npx motionscore <url> command in SKILL.md allows the agent to process data from arbitrary websites.
Boundary markers: No delimiters or instructions to ignore embedded content are provided.
Capability inventory: Subprocess execution via npm, yarn, pnpm, curl, and npx in SKILL.md.
Sanitization: There is no validation or sanitization mentioned for the input URL before it is passed to the command line utility.

motion-agent