sdmx-explorer

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

• Third-party content exposure detected (high risk: 0.80). This skill explicitly fetches and parses live content from public SDMX endpoints (e.g., via opensdmx commands like search, info, constraints, values against providers such as Eurostat, ISTAT, OECD, World Bank, ECB), and the workflow requires the agent to read that untrusted third‑party metadata/notes/codelists (including NOTE_*/flag text) to choose datasets, build queries, and drive subsequent tool actions, so external content can materially influence behavior.