astryx

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

• Third-party content exposure detected (low risk: 0.10). The runtime workflow ingests only first-party, trusted component/template/doc content from the Astryx MCP/CLI (e.g., search(query)/get(name)), and it does not ingest outsider-authored free text unless the user explicitly provides it as a specific component/template/doc query for retrieval.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill documents an MCP server that is queried at runtime by AI tools (e.g., Cursor/Claude) to fetch component/docs, so the external URL https://astryx.atmeta.com/mcp can directly control agent prompts by supplying model context at runtime.