geopolitical-risk

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The SKILL.md includes a fetch_acled_events function that accepts an api_key argument and inserts it verbatim into an HTTP request params dict, which would require the LLM (or user-facing agent) to handle and potentially output a secret API key directly.

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

• Third-party content exposure detected (high risk: 0.85). SKILL.md describes a runtime workflow that uses “read_url” to fetch real-time news from external URLs (e.g., Reuters/Bloomberg/FT/ForeignPolicy) and then “pass it to the LLM to extract risk events,” which is outsider-authored free text from public web pages.