posthog-automation

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

• Third-party content exposure detected (high risk: 0.75). The required workflow calls PostHog MCP tools (e.g., list/filter events and feature flags) which at runtime will return outsider-authored free text from the PostHog project (event names/properties and flag configuration) into the agent’s LLM context.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill instructs clients to add and use https://rube.app/mcp as the MCP server and to call RUBE_SEARCH_TOOLS (SKILL.md:11,18), meaning the agent fetches tool schemas from that endpoint at runtime which directly control tool/instruction behavior, so this is a required runtime external dependency.