arkade-product-ui

Warn

Audited by Gen Agent Trust Hub on Nov 1, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill directs the agent to clone external repositories from personal GitHub accounts if they are not already installed. Specifically, it references https://github.com/emilkowalski/skills and https://github.com/pbakaus/impeccable.
  • [REMOTE_CODE_EXECUTION]: The instructions explicitly tell the agent to "run its context command once" after cloning the external repositories. This pattern executes arbitrary commands defined by third parties within the agent's environment.
  • [INDIRECT_PROMPT_INJECTION]: The skill relies on instructions and "skill files" fetched from the cloned repositories to guide its behavior. This creates a vulnerability surface for indirect prompt injection.
  • Ingestion points: Files within emilkowalski/skills and pbakaus/impeccable repositories.
  • Boundary markers: None; the agent is instructed to read and follow the external skill files directly.
  • Capability inventory: The agent can perform git clone operations and execute "context commands" (shell execution).
  • Sanitization: There is no evidence of validation or sanitization for the content or commands fetched from these external sources.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Nov 1, 2026, 11:59 AM