dsh-workbuddy2api-hub：把本地 workbuddy2api-hub 接进 DSH
DSH 插件 dsh-workbuddy2api-hub 把本地运行的 workbuddy2api-hub 接进 DSH 桌面端，自动同步模型（带上下文窗口、图片输入、思考档位）、实时显示积分余额与用量，支持多 key 管理、自动告警与容错处理。
Read MoreDeepSeek Harness Local Privacy Gateway Plugin
The “Local Privacy Gateway” plugin for DeepSeek Harness operates under the MIT License and is designed to address compliance risks in data processing. It locally replaces sensitive information in conversations—such as names, phone numbers, and ID numbers—with placeholders, and scans CSV/XLSX files for anonymization, ensuring that data remains visible only locally and is not exposed to the model or logs. The plugin supports custom lexicons, stores data locally in `~/.dsh-privacy-gateway/`, and requires users to enable it manually and download desensitized copies.
Read Moredsh-our-free-model: Free Model Plugin for DeepSeek Harness
The DeepSeek Harness plugin `dsh-our-free-model` is designed to address API key management challenges, offering a free solution for using frontier models—such as DeepSeek V4.1 Flash and Kimi K3—without registration or the need to enter an API key. This open-source plugin features over 14,000 stars, supports zero-dependency installation, and includes thirteen free access channels. Its core features include intelligent model selection, nine-page frosted-glass interface management, local gateway forwarding, and LAN access. Please note considerations for privacy and security: reasoning intensity control affects the output budget, and API keys are stored locally only.
Read Moredsh-deepseek-web-adapter: Integrating DeepSeek Web with DSH
dsh-deepseek-web-adapter is a DSH plugin designed to access DeepSeek's web model without requiring an official API Key. Maintained by the community, it automatically starts a local gateway (default port 5688) upon loading and cleans up resources upon uninstallation, while exposing an OpenAI-compatible interface. Users must configure an arbitrary value for MOCK_LLM_KEY and manually log in to the DeepSeek web version via a browser to maintain the session state. Core features include continuous conversation, tool calling, and model calibration, supporting DeepSeek's Quick, Expert, and Vision modes. While login, basic Q&A, and single tool calling have been verified, capabilities such as sub-agent concurrency, multi-round tool loops, and long-context conversation migration have not been fully tested. This is currently a developer version.
Read More