twg-confluence

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

• Third-party content exposure detected (high risk: 0.75). Confluence content is fetched at runtime via outsider-provided anchors (e.g., a Confluence URL/content ID/CQL), and the skill then reads the selected page body/snippets into the agent context (e.g., confluence content get / search → native get), which can include free text authored by non-operating users.