draft-message

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill prompt embeds literal Telegram API credentials (API_ID and API_HASH) in example commands and code, which would require the LLM to reproduce those secret values verbatim in outputs/commands, creating an exfiltration risk.

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

• Secret detected (high risk: 1.00). I flagged a real secret: the Telethon API hash appears as a literal high-entropy string in two code examples:
• "4b812166a74fbd4eaadf5c4c1c855926"

This matches the definition of a secret (long, random-looking API credential) and can be used with the accompanying API_ID (18256514) to authenticate to Telegram. It is not a placeholder, truncated value, or a simple example password.

Ignored items / rationale:

• The numeric API_ID (18256514) is present but is not high-entropy on its own; however it is used with the API_HASH and together they form usable credentials.
• Other strings (session path, chat IDs like -1003958083153, example labels, or sample messages) are non-sensitive identifiers or documentation text and were not flagged.