file-browser

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (medium risk: 0.30). The visible skill is a local file viewer with no malicious intent, but reader.js weakens protections by setting Mermaid's securityLevel to 'loose' and inserting unescaped error and source text into innerHTML, which is careless and risky toward the user's rendered content.

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

• Third-party content exposure detected (low risk: 0.10). The runtime path serves user-supplied local files and directory listings (/view, /browse, /api/tree, /api/search) that the agent points at with --file/--dir, and these are not outsider-submitted feeds, so exposure is at most low, from selected local content that may include third-party files.