DeepSeek Harness Plugins · Workflow
Plugins that extend DeepSeek Harness (DSH) with vision, workflows, memory, web tools and more.
A DeepSeek Harness plugin that registers a global skill to guide agents in discovering, evaluating, and installing community plugins, avoiding reliance on web search.
The dsh-timeline-studio-plugin connects DeepSeek Harness to Timeline Studio, enabling agents to inspect, edit, preview, and render video projects through natural language with secure workflow tools.
DSH Notification Framework that unifies notification event management, supports derived plugin registration and custom configuration, with multiple notification channels and sound settings.
Preserves minimal mode context boundaries with on-demand vision capabilities, supporting image and document processing.
A plugin for DeepSeek Harness that ports Matt Pocock's 25 engineering and productivity skills to DSH's Cordis architecture, automatically registering them for agent use.
Adds a visual agent team to DeepSeek Harness, enabling member communication, task claiming, shared workspace, and a 2.5D office collaboration interface.
Turn your devices into a collaborative fleet with auto-discovery, pairing, and remote execution, all as a zero-core-change plugin.
A DSH plugin that automatically notifies the agent after a service restart, allowing session recovery without user prompts.
Unifies plugin creation, diagnosis, testing, and upgrade verification into a single workflow within DSH or Codex, from ideation to delivery.
Spawns background Shadow agents after eligible tool turns, validates their structured findings, relays accepted reports into the root session, and exposes lifecycle logs and debugging tools in the Web UI.
A scheduled loop plugin that provides the /loop command and loop tool for self-adjusting cycles, supporting multiple concurrent loops in agent workflows.
dsh-orchestrator is a plugin for DeepSeek Harness that enables task decomposition and dispatch via a main agent, direct messaging between any agents, and per-worker model and thinking intensity configuration.