confluence-update
Pass
Audited by Gen Agent Trust Hub on May 19, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The script launches a Chromium-based browser (Chrome, Chromium, Brave, Edge, or Vivaldi) using child_process.spawn. It enables remote debugging via the --remote-debugging-port flag to allow programmatic interaction with the browser session for authentication purposes.\n- [DATA_EXFILTRATION]: The skill retrieves Atlassian-specific session cookies from the browser via a WebSocket connection to the Chrome DevTools Protocol (Network.getCookies). These credentials are used exclusively to authenticate REST API requests to the user-specified Atlassian Confluence site.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes external data to update Confluence pages, creating a potential surface for indirect injection if an attacker controls the input files.\n
- Ingestion points: Content is read from local files specified by the --file argument in scripts/confluence-update.js.\n
- Boundary markers: The replace-block command uses specific HTML comment delimiters () to scope updates.\n
- Capability inventory: The skill can create, update, and replace content on Confluence pages via Atlassian's REST API using fetch calls in scripts/atlassian-browser.js.\n
- Sanitization: Content is converted from Markdown to Confluence Storage XHTML using markdownToStorage in scripts/lib.js, which includes HTML escaping for code blocks and basic text elements.
Audit Metadata