google-finance

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). Insecure: the prompt includes literal cookie values (CONSENT and SOCS) and instructs the script to pass them automatically, which requires embedding those secret cookie strings verbatim in requests/code and thus poses an exfiltration risk.

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

• Third-party content exposure detected (high risk: 0.85). El skill llama a un endpoint interno de Google (POST .../FinHubUi/data/batchexecute) y luego parsea el wrb.fr response como JSON doble; ese payload incluye texto de terceros (p.ej. titulares/URLs de news y descripciones/metadata de empresas) que se vuelca al contexto/outputs del agente como texto legible.

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

• Secret detected (high risk: 1.00). I scanned the entire skill docs and assets for literal high-entropy credentials.

Findings:

• The SOCS cookie value ("CAESHAgBEhJnd3NfMjAyNTAxMjMtMF9SQzMaAmVuIAEaBgiAvLW8Bg") appears verbatim in multiple places (scripts and assets/consent_cookies.json). It is a long, random-looking token used as a consent-bypass cookie — i.e., a cookie value that would grant access/behavior change on google.com. This meets the definition of a secret (high-entropy literal that grants access).
• The CONSENT value ("PENDING+999") is short and non-random and is a documented placeholder/standard bypass string; per the rules this is a documentation/setup value and should be ignored.
• Other values (RPC IDs, BL_VERSION build strings, User-Agent, example commands, tickers, etc.) are not secrets (either identifiers, low-entropy, or documentation examples) and are ignored.

Therefore I flag the SOCS cookie value as a real secret present in the documentation. The CONSENT value is ignored as a non-secret placeholder.

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

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