extension-guide-v3

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

• Third-party content exposure detected (high risk: 0.90). The SKILL.md explicitly instructs the agent to perform WebSearch and call GitHub APIs (e.g., "Research Latest Versions" using WebSearch and "gh api repos/.../releases/latest") and to ingest releases, npm/apt/ direct URLs (api.github.com, registry.npmjs.org, apt repo URLs) to determine and update versions — i.e., it fetches and interprets untrusted public web content that can change install/upgrade actions.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill includes explicit runtime commands that fetch and execute remote packages (e.g., "npx -y @mytool/mcp-server start" which pulls code from the npm registry / registry.npmjs.org, and binary downloads referencing https://github.com/org/repo), so external content can be fetched and run at runtime.