genshijin-compress

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The skill sends raw file contents to an external API and mandates verbatim preservation of code blocks/inline code (and other literals), so any secret values embedded in those files would be transmitted and output unchanged (the refusal of known secret filenames reduces but does not eliminate this risk).

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

• Potentially malicious external URL detected (high risk: 0.90). The skill's call_claude function sends file contents at runtime to the Anthropic API (via the anthropic client / implicit endpoint https://api.anthropic.com or the claude CLI) and uses the returned model output to overwrite files, so a remote URL is contacted during runtime and the fetched content directly controls the agent's resulting file output.