nlweb-schema-org-grounding

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

• Third-party content exposure detected (high risk: 0.85). SKILL.md instructs runtime ingestion via “URL-list ingest path extracts” Schema.org JSON-LD from fetched third-party web pages (e.g., <script type="application/ld+json"> on arbitrary URLs), which becomes readable schema_object text/JSON in the agent’s LLM context.

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 live files that define per-type prompts and tool inheritance — https://github.com/nlweb-ai/NLWeb/blob/main/config/site_types.xml and https://github.com/nlweb-ai/NLWeb/blob/main/docs/nlweb-prompts.md — which are fetched at runtime and directly control the agent's prompts/instructions.