currents-agent-skills

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (medium risk: 0.30). The skill includes explicit instructions for the agent to create and overwrite local report files (including "Do not ask the user for permission to create it"), which weakens user control and constitutes suspicious steering without clearly malicious intent.

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

• Third-party content exposure detected (medium risk: 0.30). The required runtime workflow ingests error text (messages, stack traces) from the Currents MCP server via tools like currents-get-spec-instance, currents-get-errors-explorer, and currents-get-test-results, whose contents ultimately originate from CI test outputs authored by the user/tenant (outsider-controlled in practice via committing tests).