design-md

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

• Third-party content exposure detected (medium risk: 0.65). The required runtime workflow includes reading the project’s DESIGN.md (or docs/DESIGN.md) in full, which is outsider-authored free text if the operating user did not author that file; this prose/tokens are then injected into the agent’s context via the “Consumption Flow”/“Consume mode” steps.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill instructs running remote code via npx (e.g., npx @google/design.md@latest spec), which fetches and executes a package from the npm registry at runtime and whose "spec" output is explicitly injected into agent prompts (also references the upstream repo URL https://github.com/google-labs-code/design.md), so external content can directly control agent instructions.