zip-unzip-file-compression-10mb

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). This skill accepts arbitrary base64-encoded file contents in request fields (files.input_base64 / input_base64) and can return those contents inline via include_contents, so an LLM composing tool calls may be required to include secret data verbatim (exfiltration risk) instead of referencing stored file IDs.

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

• Third-party content exposure detected (medium risk: 0.65). The runtime path is the decompress action taking input_base64 or file_id (previously stored archive) and then returning extracted file contents (optionally via include_contents), meaning any outsider-authored archive contents can be ingested as readable base64 text into the agent’s LLM context.