pp-human-goat

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

• Third-party content exposure detected (medium risk: 0.65). The required workflow is to run the CLI with --agent, which returns JSON that the agent will read into its LLM context; that JSON can include outsider-authored free text from TaskRabbit/Magic (e.g., tasker/review/message content) and the skill does not sanitize it before LLM ingestion.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.80). The skill requires installing the CLI via remote Go modules which, if executed at runtime, fetch and build remote code from github.com/mvanhorn/printing-press-library/library/productivity/human-goat/cmd/human-goat-pp-cli@latest and github.com/mvanhorn/printing-press-library/library/productivity/human-goat/cmd/human-goat-pp-mcp@latest, thereby executing remote code as a required dependency.

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

• Direct money access detected (high risk: 1.00). The skill exposes explicit, non-generic payment/checkout actions: commands like "hire" perform hands-off checkout "against the card on file" and will book (charge) a TaskRabbit task unless capped; "cancel" reverses bookings and reports fees. This is an explicit ability to create payment transactions (autonomous checkout), not a generic browser or HTTP tool. Therefore it grants direct financial execution authority.