gstack

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

• Third-party content exposure detected (high risk: 0.90). The SKILL.md and BROWSER.md explicitly let the agent navigate to and read arbitrary external URLs (commands like goto <url>, text, html, links, snapshot, diff shown in SKILL.md/BROWSER.md), meaning the agent fetches and interprets untrusted public web content as part of its workflow and can take subsequent actions based on that content.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill's setup flow may run a one-time installer that executes remote code via "curl -fsSL https://bun.sh/install | bash" (used during runtime if Bun is missing), so https://bun.sh/install is a runtime-fetched URL that would execute remote code as a required setup dependency.