BenjaminSHI4008/deepseek-pet-Seeki ↗★ 2
deepseek-pet
DeepSeek Harness 桌宠插件：透明置顶的桌面精灵（2D 像素风状态机）
AI 分析
核心用途是提供一个趣味性的桌面像素风宠物，并集成 DSH 聊天窗口。适合想要在桌面放置互动挂件并快速发起 AI 对话的用户。需要 Node.js >= 22。
インストール
検証済み bundle がないか、互換性チェックに失敗しています。先にリポジトリの説明を読んでください。 README 全文を読む ↗
ドキュメント
README 全文を読む ↗Model Configuration
Seeki doesn't hardcode models. The chat window's model list comes from your DeepSeek Harness's configured providers (llm.models), so it always reflects what your harness actually advertises (for example DeepSeek-V4-Pro and DeepSeek-V4-Flash on the deepseek-official route).
- API key — configure it for the harness, not for Seeki: set
DEEPSEEK_API_KEYin the environment, in a.envfile at the harness root, or via the harness's credentials store (~/.dsh/.credentials.yaml). - Default model — the harness's own default. Seeki doesn't impose one.
- Switch models — use the model dropdown in the chat title bar. It affects the next message only; new conversations inherit the current selection; unavailable models show as disabled.
Full pet.config.json reference: docs/config.md.