asana-automation

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

• Third-party content exposure detected (medium risk: 0.65). The required workflow is driven by Composio/Rube MCP tool calls (e.g., Asana search/list/get), which at runtime can return outsider-authored free text from Asana objects (task names/notes/descriptions) into the agent’s LLM context via tool results; SKILL.md itself does not show sanitization or filtering.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill requires adding the MCP endpoint https://rube.app/mcp (SKILL.md line 15) which the agent will call at runtime (e.g., via RUBE_SEARCH_TOOLS) to fetch tool schemas/instructions that directly control agent behavior, making it a required remote instruction source.