parallel-cli

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

• Third-party content exposure detected (medium risk: 0.30). Runtime path: parallel-cli search/research run/extract read outsider-authored free text from URLs determined by the user-supplied OBJECTIVE/URLs, which can include attacker-posted content; this free text is ingested into LLM context via the CLI’s fetch/extract workflow.

MEDIUM W009: Direct money access capability detected (payment gateways, crypto, banking).

• Direct money access detected (high risk: 1.00). The skill explicitly includes commands to check and add account funds via the Parallel CLI (e.g., parallel-cli balance --json get and parallel-cli balance add <AMOUNT_IN_CENTS>). That balance add command is a specific financial execution action (topping up account funds), so this skill grants direct financial execution capability.