linux-ai-dev-stack

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.70). The skill contains instructions and code examples that execute curl commands to fetch and run remote installation scripts from third-party URLs (such as ollama.com/install.sh, claude.ai/install.sh, chatgpt.com/codex/install.sh, and opencode.ai/install), which constitute unverified runtime external dependencies with weak or unknown provenance.