init-claude

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

• Third-party content exposure detected (medium risk: 0.30). このスキル（init-claude）は実行時に「対象リポジトリのファイル（package.json/Cargo.toml/pyproject.toml/Makefile/.github/workflows 等）」を cat/ls/find で読み取り、その内容を Step 2 の構成案設計に利用します（外部投稿文を直接検索・特定しない前提でもリポ内テキストがLLMに渡り得るため）。

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

• Potentially malicious external URL detected (high risk: 0.90). The skill runs "npx skills add Fandhe-AI/agent-cli-skills" at runtime to fetch and install external skill code (providing files like ../skills/implement-issue-tree/scripts/implement-issue-tree.js) which the workflow and agent behavior rely on, so remote code is fetched and executed and controls agent functionality.