milady-development

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly instructs fetching and syncing plugins from upstream git remotes (e.g., "git fetch origin next"/"git pull origin next") and installing/ejecting registry plugins (e.g., "Install @elizaos/plugin-twitter", "Eject plugin-discord") so the agent will ingest third‑party plugin code (read via reading index.ts in ~/.milady/plugins/) which can alter its actions and thus enable indirect prompt injection.

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

• Potentially malicious external URL detected (high risk: 0.70). The skill includes runtime commands that fetch and install code from remote git origins (e.g., "cd ~/.milady/plugins && git pull origin next && git fetch origin next && bun install"), which would load plugin code that can change agent prompts and execute behavior—so the remote git "origin" repositories referenced by those git pull/fetch commands are flagged.