quickjs-sandbox

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

• Third-party content exposure detected (high risk: 0.85). The required workflow spawns a “research sub-agent” that fetches documentation pages from https://sebastianwessel.github.io/quickjs/... at runtime; this is public web content (outsider-authored) that could be ingested as free text into the agent’s LLM context via the sub-agent’s web fetching path.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly instructs a spawned sub-agent to fetch documentation pages from https://sebastianwessel.github.io/quickjs/ at runtime and to use the fetched content to drive prompts, answers, and code examples, so that URL is a runtime external dependency that directly controls agent output.