dsh-tool-ocr
@ferstar
Multimodal
Local OCR plugin that enables text-only LLMs to read images using the newbee-ocr engine over PP-OCRv6 models, with settings card and attachment support.
DeepSeek Harness Plugin Marketplace
1574 pluginsGitHub data updated Aug 28, 2026
dsh-sysmon
@AKS1st
Monitoring
System-status overlay for DSH Web showing live CPU, memory and disk usage, refreshed every second with color-coded thresholds.
dsh-zh-hant-hk
@Argonaut790
UI & Experience
Adds Hong Kong Traditional Chinese wording to the DeepSeek Harness UI, converting Simplified Chinese chrome and applying local terminology.
dsh-file-panel
@yu2025-luo
UI & Experience
Right-side file panel for DeepSeek Harness that auto-pops up when the agent creates or downloads files, with preview, reveal-in-folder, and open actions.
dsh-docker
@STARDUSTLC666
Coding Tools
Container management plugin for DeepSeek Harness agents, providing docker_ps/logs/inspect/exec/manage tools via the official subprocess service with argv-based no-shell injection protection, exec approval gate, and zero runtime dependencies.
dsh-plugin-vm-sandbox
@GHJIVHIDD
Security
VM sandbox plugin for DeepSeek Harness: provides per-session OrbStack VMs (debian/alpine) with a 'Virtual Machines' tab, tools for VM management, and resource governance.
dsh-memory
@Max-Null
Agent & Workflow
Cross-session plaintext memory for DeepSeek Harness: suggested → human-approved, searchable, human owns the data · 跨会话明文记忆：模型写入待审核、人工确认生效，明文可审计
dsh-open-eyes
@hyper-dsh-plugins
Multimodal
Adds vision capabilities to DeepSeek Harness by delegating image analysis to external multimodal models via OpenAI or Anthropic adapters.
dsh-memory-protocol
@baaai123
Agent & Workflow
Memory enforcement plugin for DeepSeek Harness that bridges the opencode-memory MCP server with a hard protocol gate, enforcing memory weaving before tool calls and auto-archiving each turn.
DHS-multi-agent-plugin
@Li3NGa
Agent & Workflow
Multi-agent orchestration runtime for DeepSeek Harness, defining agent teams with structured task graphs (DAG) and six collaboration strategies, usable via Python API, CLI, HTTP, or MCP.
