agent-package-manager

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

• Third-party content exposure detected (high risk: 0.75). Outsider-authored free text can enter the LLM context when the required runtime workflow runs apm install/apm deps update to fetch third-party packages/skills (e.g., owner/repo/path), because those fetched repositories’ prompt/agent/skill markdown files are ingested as readable text into the agent’s deployed context.

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

• Potentially malicious external URL detected (high risk: 0.90). The troubleshooting steps explicitly instruct running remote installers that are fetched-and-executed at runtime (curl -sSL https://aka.ms/apm-unix | sh and irm https://aka.ms/apm-windows | iex) to install the required apm CLI, which is remote code execution used to enable the skill workflow.