build-it

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

• Third-party content exposure detected (high risk: 1.00). The skill's Phase 3 workflow explicitly fetches GitHub issues using "gh issue view --json body,title", which pulls untrusted, user-generated content (public GitHub issue bodies) that the agent must parse and act on to drive implementation decisions and code generation.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill explicitly fetches a GitHub issue at runtime using "gh issue view " (i.e. https://github.com/.../issues/), and it parses the issue body as the specification that directly controls the agent's instructions, so this remote content can directly control the agent at runtime.