github

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (medium risk: 0.65). These links point to an unfamiliar vendor (deepnlp.org) plus a GitHub repo and Product Hunt listing that instruct installing npm/pip packages and running scripts — there are no direct .exe downloads or URL shorteners, but installing/running packages from an unestablished publisher and running provided scripts carries a non-trivial risk of supply‑chain or arbitrary code execution.

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

• Third-party content exposure detected (high risk: 0.90). This skill calls GitHub via the OneKey Gateway (e.g., SKILL.md tools like get_file_contents, pull_request_read, search_code and the corresponding scripts such as scripts/get_file_contents.py) to fetch public repository files, PR diffs, issues and comments (user-generated, untrusted content) which the agent is expected to read and which can directly influence actions like create_or_update_file or merge_pull_request (the SKILL.md even notes automatic fetching of SHAs), so third-party content can alter tool use and decisions.