linkfox-temu-fulfillment-eu
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill exposes a data ingestion surface through external Temu API responses that are directly printed to standard output or saved as files within the workspace.
- Ingestion points: Multiple execution entry points including
scripts/temu_eu_proxy.pyand specific domain scripts ingest external data directly from the Temu gateway (/temu/proxy). - Boundary markers: No specific delimiters or boundary markers are placed around the parsed JSON data fields before they are printed to
stdoutand swallowed by the agent context. - Capability inventory: The utility scripts possess capabilities to perform outbound network requests (
urllib.request.urlopen) and file creation within the workspace folder structure (linkfox/<date>/<session>/data/). - Sanitization: The tool outputs raw text and structured content directly from upstream servers without sanitizing potential natural language commands embedded in fields like item names or descriptions, posing a tool output poisoning surface.
Audit Metadata