alibabacloud-solution-deploy

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly asks users to provide passwords/API keys (e.g., "Password: (请提供)"), shows deployment parameters that include sensitive values, and instructs generating CLI commands and Terraform files using those parameters—forcing the agent to handle and potentially output secrets verbatim.

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly instructs the agent to fetch and parse third‑party deployment examples (Step A.2: "Fetch the example main.tf via WebFetch" with raw GitHub URLs) and to fetch public solution tutorial pages (Step B.1), which the agent will read and use as defaults that can materially change deployed CLI/terraform commands, exposing it to untrusted user-generated web content.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly fetches example Terraform code at runtime from raw GitHub URLs (e.g. https://raw.githubusercontent.com/alibabacloud-automation/terraform-alicloud-/main/examples/complete/main.tf), and that fetched HCL is used as deployment input which directly controls/executed infrastructure changes via the IaCService, so this is a high-confidence runtime external dependency that can execute remote code.