confluence-api

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

• Third-party content exposure detected (high risk: 0.85). Outsider-authored Confluence page content is fetched at runtime via scripts/read_page.sh (GET /api/v2/pages/{id}?body-format=view|storage|export_view) and the returned body.*.value is printed to stdout (and thus becomes LLM-readable context); Confluence pages/comments can contain adversarial instructions from other authors.

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

• Potentially malicious external URL detected (high risk: 0.90). This skill performs runtime requests to the Confluence site (CONFLUENCE_BASE, e.g. https://your-domain.atlassian.net/wiki and its /api/v2 and /rest/api/ endpoints) to fetch page bodies/attachments which are returned for use by the agent — such fetched content can contain adversarial instructions and thus can directly control prompts if ingested, so it is a runtime dependency that poses prompt-injection risk.