mastering-confluence

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly downloads and ingests arbitrary Confluence pages and attachments using MCP tools and scripts (e.g., download_confluence.py, mcp__atlassian__confluence_search and confluence_get_page) — these are user-generated/untrusted third-party contents that the agent is expected to read and convert, so they could carry indirect prompt-injection payloads.