DeepSeek Unveils First Agent Framework, Harnessing AI with Plug-and-Play Execution Engine
Following its recent API pricing updates, DeepSeek unveiled DeepSeek Harness v0.1 Developer Preview on August 13, open-sourcing the project on GitHub under the MIT license. This initiative, championed by Liang Wenfeng, shifts focus from the model itself to the supporting infrastructure that enables AI to execute real-world tasks. The team defines this relationship with a simple formula: Model + Harness = Agent, where the model handles reasoning and Harness manages execution.

Harness provides a comprehensive local agent workspace, featuring project management, long-term task collaboration, multi-agent orchestration, context handling, online retrieval, and external skill integration. This allows developers to run complex tasks immediately without assembling disparate tools. As an early preview, Harness operates on the principle that "Everything is a plugin," redefining Agent development through an open architecture.
Four Operational Modes, Configured via Assembly
The core strength of Harness is its plugin-based open architecture. Every component—models, tools, skills, sessions, sandboxes, storage, loops, scheduling, and UI—is a plugin that can be swapped or reassembled freely. Built on the Cordis plugin meta-framework, Harness handles only low-level tasks like loading and dependency management. All business logic consists of independent Cordis plugins that interact via services and events, allowing developers to reconfigure the system at the configuration level without altering source code.
Harness supports four predefined operational modes, each loading a specific plugin set: Standard mode includes a full toolset for general development; PTC (Programmatic Tool Call) mode enables the model to generate code and orchestrate multi-turn tool chains; Minimal mode restricts tools to Shell and file editing for baseline testing in constrained environments; and Creative mode offers runtime status viewing, memory debugging plugins, and support for custom operational modes. Developers can quickly launch a Web interface locally using the npx command after deploying a Node.js toolchain.
As DeepSeek’s foundation for the Agent arena, v0.1 is still in its early stages, yet its "disassemble-and-reconfigure" philosophy addresses current fragmentation in the intelligent agent ecosystem. While competitors like OpenAI push for Agent Plugins standards and vie for influence, DeepSeek’s decision to open the entire stack under the MIT license lowers barriers for developers building custom Agents, significantly boosting the domestic open-source Agent ecosystem.
Related article
Swedish AI Startup Lovable Eyes $13.2 Billion Valuation After Major Funding Round
As AI-driven coding tools gain traction, Swedish startup Lovable has secured a major funding round. The company aims to raise $3 billion, potentially boosting its valuation to $13.2 billion—double the $6.6 billion recorded last December. Menlo Ventur
Google Tests Remy AI Agent for Gemini as Focus Shifts to User Control
According to Business Insider, Google is testing Remy, a new AI personal agent for Gemini. This tool aims to execute tasks on behalf of users, streamlining both professional workflows and daily routines.Currently, Remy is undergoing testing in an int
How to fix Core Web Vitals for better SEO rankings
Streamline Report Card Comments with AI ToolsIntroductionAI Tools for Generating Report Card CommentsMagic SchoolAlmanac AIChat GPTUsing Magic School to Generate Report Card CommentsLogging into Magic SchoolSelecting the Report Card Comments ToolCust
Related Special Topic Recommendations
Comments (0)
0/500
Following its recent API pricing updates, DeepSeek unveiled DeepSeek Harness v0.1 Developer Preview on August 13, open-sourcing the project on GitHub under the MIT license. This initiative, championed by Liang Wenfeng, shifts focus from the model itself to the supporting infrastructure that enables AI to execute real-world tasks. The team defines this relationship with a simple formula: Model + Harness = Agent, where the model handles reasoning and Harness manages execution.

Harness provides a comprehensive local agent workspace, featuring project management, long-term task collaboration, multi-agent orchestration, context handling, online retrieval, and external skill integration. This allows developers to run complex tasks immediately without assembling disparate tools. As an early preview, Harness operates on the principle that "Everything is a plugin," redefining Agent development through an open architecture.
Four Operational Modes, Configured via Assembly
The core strength of Harness is its plugin-based open architecture. Every component—models, tools, skills, sessions, sandboxes, storage, loops, scheduling, and UI—is a plugin that can be swapped or reassembled freely. Built on the Cordis plugin meta-framework, Harness handles only low-level tasks like loading and dependency management. All business logic consists of independent Cordis plugins that interact via services and events, allowing developers to reconfigure the system at the configuration level without altering source code.
Harness supports four predefined operational modes, each loading a specific plugin set: Standard mode includes a full toolset for general development; PTC (Programmatic Tool Call) mode enables the model to generate code and orchestrate multi-turn tool chains; Minimal mode restricts tools to Shell and file editing for baseline testing in constrained environments; and Creative mode offers runtime status viewing, memory debugging plugins, and support for custom operational modes. Developers can quickly launch a Web interface locally using the npx command after deploying a Node.js toolchain.
As DeepSeek’s foundation for the Agent arena, v0.1 is still in its early stages, yet its "disassemble-and-reconfigure" philosophy addresses current fragmentation in the intelligent agent ecosystem. While competitors like OpenAI push for Agent Plugins standards and vie for influence, DeepSeek’s decision to open the entire stack under the MIT license lowers barriers for developers building custom Agents, significantly boosting the domestic open-source Agent ecosystem.
Swedish AI Startup Lovable Eyes $13.2 Billion Valuation After Major Funding Round
As AI-driven coding tools gain traction, Swedish startup Lovable has secured a major funding round. The company aims to raise $3 billion, potentially boosting its valuation to $13.2 billion—double the $6.6 billion recorded last December. Menlo Ventur
How to fix Core Web Vitals for better SEO rankings
Streamline Report Card Comments with AI ToolsIntroductionAI Tools for Generating Report Card CommentsMagic SchoolAlmanac AIChat GPTUsing Magic School to Generate Report Card CommentsLogging into Magic SchoolSelecting the Report Card Comments ToolCust





Home






