Hey Product Hunt 👋 I built SpecsRelay because I often use DeepSeek to explore a product idea, compare options, and confirm decisions—but implementation continues inside a local DSH project. The painful part was the handoff: copying a long conversation, rebuilding context, separating confirmed decisions from suggestions, and pointing the Agent at the right project. SpecsRelay puts that handoff inside supported DSH desktop clients. It opens the real DeepSeek website, captures the current complete conversation only when I click Organize current conversation, uses the model already configured in DSH to form an actionable requirement, asks for clarification only when a missing decision would affect implementation, and sends the confirmed result to the selected project to start the Agent. It is open source under MIT. There is no browser extension, Docker service, extra platform, or second API key in this workflow. The project is independently maintained and is not an official DeepSeek product. The current release supports four community desktop clients: DSH Desktop by anywhere-labs, Pilot Harness, DataElement DSH Desktop, and myYangyunfan DSH Desktop. The anywhere-labs client is the recommended option for regular users; complete support for the other three currently uses SpecsRelay companion builds. I would love feedback on two things: where does your AI discussion-to-implementation handoff break today, and which DSH desktop workflow should SpecsRelay support next?

SpecsRelay for DeepSeek

Turn DeepSeek conversations into work your DSH Agent can run

Turn DeepSeek conversations into work your DSH Agent can run