vercel-agent-writing-guidelines

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

• Third-party content exposure detected (high risk: 0.85). The skill’s required workflow uses WebFetch to retrieve free-form guideline text from a public URL at runtime (outsider-authored web content), which is then ingested into the agent’s LLM context for rule application.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly instructs fetching runtime rules from https://raw.githubusercontent.com/vercel-labs/writing-guidelines/main/command.md and uses that fetched content to control the agent's checking logic and output format, making it a required external dependency that directly controls prompts/instructions.