sync-trending

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 1.00). The skill fetches and analyzes content from public, user-generated sources—primarily GitHub Trending (https://github.com/trending) and cloned repositories—then reads/executes their READMEs and code as part of its workflow, exposing the agent to untrusted third-party content that could contain indirect prompt injection.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill fetches live content from https://github.com/trending at runtime (via web_fetch) and, with user permission, uses git clone to pull and install/run those trending repositories—thereby retrieving and potentially executing arbitrary remote code from those URLs.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 0.90). The skill instructs the agent to clone third‑party repositories and autonomously install and run their code on the host (even in a “temporary” dir), which modifies the machine's state and can execute arbitrary/privileged operations—so it significantly pushes the agent toward compromising the host despite asking for permission.