morningstar

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). El prompt contiene un token explícito ("klr5zyak8x") que el agent/skill usa directamente en requests/endpoints, lo que obliga a manejar y potencialmente reproducir ese secreto literalmente en salidas (alto riesgo de exfiltración).

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

• Third-party content exposure detected (high risk: 0.85). En los modos search y screener, el script hace requests a un endpoint público de Morningstar (tools.morningstar.{co.uk,de,fr,it,es}/api/rest.svc/.../security/screener), parsea la respuesta JSON y la inserta como texto/JSON en la salida del programa (y por tanto en cualquier contexto LLM que consuma esa salida); esa respuesta proviene de un tercero (Morningstar) y no del usuario.

HIGH W008: Secret detected in skill content (API keys, tokens, passwords).

• Secret detected (high risk: 1.00). The file includes a literal token value "klr5zyak8x" in multiple places:
• TOKEN = "klr5zyak8x" in scripts/fetch_morningstar.py
• Used in endpoint URL patterns (e.g. /api/rest.svc/klr5zyak8x/security/screener) and called "token universal" in REFERENCE.md / UNIVERSES.md / SKILL.md.

This is not a placeholder (e.g., YOUR_API_KEY) nor a simple setup password; it is an actual token embedded in code and used to access the internal API endpoint across subdomains. Even though the string is relatively short, it functions as an access credential and therefore qualifies as a secret per the definition ("literal value that provides access to a service"). No other high-entropy API keys, private keys, or passwords are present. PerformanceId values and universe codes are identifiers, not credentials, so they are ignored.

MEDIUM W021: Hidden or invisible Unicode characters detected (potential obfuscation or prompt injection).

• Hidden Unicode characters detected (1 type(s) found)