paperclip-monitor

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

• Third-party content exposure detected (medium risk: 0.30). The skill’s runtime workflow for paperclip-monitor reads Paperclip control-plane records such as issue lists/details, comments, approvals, and especially issue execution data (paperclipai issue live-runs, active-run, runs, recovery-actions), which can include outsider-authored free text from issue/comment bodies submitted by non-trusted users.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill instructs running "npx -y @bbengamin/paperclip-mcp-server" as an MCP fallback (references/docs/...); npx fetches and executes remote package code at runtime, which is an external dependency that will run code locally.