nfeio-node-sdk

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The prompt includes explicit examples that hard-code API keys and webhook secrets (e.g., apiKey: 'your-api-key', secret: 'a-secret-with-32-to-64-characters-x') and shows using the webhook.secret in signature validation, which can lead an LLM to emit or propagate real secret values verbatim rather than using safe env/CLI-based patterns.

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

• Third-party content exposure detected (high risk: 0.75). Outsider free text can enter the LLM context via webhook payloads: nfe.webhooks.validateSignature(req.body, ...) implies the agent will parse and use req.body (webhook event JSON) from an external sender at runtime, which is not authored by the operating user.

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

• Secret detected (high risk: 1.00). I scanned the entire skill content for literal credentials. Most credential-like strings are clearly placeholders or examples and are ignored per the rules:
• Ignored placeholders: 'your-key', 'your-api-key', 'YOUR_API_KEY', process.env usages, and similar environment-variable names.
• Ignored human-readable example secret: 'a-secret-with-32-to-64-characters-x' (obvious sample text, low entropy).
• Ignored short/simple numeric examples such as CNPJ/CPF examples (e.g., 11444555000149, 12345678000190) — these are tax identifiers/sample data, not credentials to external services.
• No PEM/private-key blocks or redaction markers like REDACTED_SECRET_ were present.

Flagged item:

• The document contains a literal 44-digit access key: '35210512345678000190550010000001231123456789' (used in productInvoiceQuery.retrieve examples). The skill itself states access keys "must be exactly 44 numeric digits" and this literal matches that format. Because it is a full-length, specific access key (a credential format used to retrieve invoices), it qualifies as a potentially usable secret and should be treated as a leaked credential.

Therefore I mark the presence of a real high-entropy credential (the 44-digit access key).