skill-portability

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly instructs spawned subagents to WebFetch the provider docUrls listed in scripts/providers.mjs (e.g., https://docs.claude.com/..., https://docs.cursor.com/..., GitHub/agentskills.io) as part of its required audit workflow (see SKILL.md step 3 and references/audit-prompt.md), so it ingests untrusted public third‑party pages whose content can change the agents' verdicts and subsequent actions.

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

• Potentially malicious external URL detected (high risk: 0.70). The skill runtime instructs subagents to WebFetch the provider doc URLs listed in scripts/providers.mjs (for example, https://docs.claude.com/en/docs/claude-code/skills) and inject those documents into the subagent prompts to guide the audit, so remote content can directly control model instructions.