Fulmar is a native Mac app that wraps the DeepSeek Harness (DSH) agent runtime, which handles conversations, tools and tasks. It adds management controls the raw runtime lacks: choosing local or cloud model routes with explicit consent, Keychain credential storage, workspace/tool access limits, and recovery checkpoints. It targets developers running file- and command-capable agents on macOS who want oversight and safety.
Visit project ↗DeepSeek Harness link shared on Hacker News
DeepSeek Harness link shared on Hacker News
DeepSeek released harness 0.2, reorganizing the agent runtime around a Profile + Bundle model and adding experimental async user interaction. Windows and macOS desktop clients are now official, with account login hinting at paid plans. The release focuses on composability, permission boundaries, and recovery paths for production use.
DeepSeek released harness 0.2, reorganizing the agent runtime around a Profile + Bundle model and adding experimental async user interaction. Windows and macOS desktop clients are now official, with account login hinting at paid plans. The release focuses on composability, permission boundaries, and recovery paths for production use.
Users exchange recommendations for DeepSeek harness skills and tools
Users exchange recommendations for DeepSeek harness skills and tools
A movo developer reports the small team abandoned its self-built agent runtime after DeepSeek released Harness, rebuilding movo on top of it. The team now focuses on agent management — access control, tool-call auditing, and knowledge-base isolation — rather than runtime engineering, and open-sourced its management layer.
A movo developer reports the small team abandoned its self-built agent runtime after DeepSeek released Harness, rebuilding movo on top of it. The team now focuses on agent management — access control, tool-call auditing, and knowledge-base isolation — rather than runtime engineering, and open-sourced its management layer.
The dsh-sieve GitHub project introduces context engineering and token optimization plugins for DeepSeek Harness (DSH). It filters tool output, prunes older context, and progressively discloses skills to reduce redundant input tokens in AI coding agents. Benchmarks report large payload reductions with archived originals retrievable on demand.
The dsh-sieve GitHub project introduces context engineering and token optimization plugins for DeepSeek Harness (DSH). It filters tool output, prunes older context, and progressively discloses skills to reduce redundant input tokens in AI coding agents. Benchmarks report large payload reductions with archived originals retrievable on demand.