DeepSeek Harness プラグイン — 247 ページ

Vision-augmented DeepSeek adapter plugin for DeepSeek Harness: a vision-capable model describes image input, then a text-only DeepSeek model reasons over the description

OpenAI-compatible local model adapter (Ollama, llama.cpp, Foundry Local, LM Studio, vLLM, gateways) for the DeepSeek Harness LLM seam

本地 LM Studio 视觉识别插件：连接 LM Studio OpenAI 兼容接口（默认 http://127.0.0.1:1234），用视觉模型（自动使用最后加载的模型，不预设具体模型）识别本地图片与剪贴板图片。agent 工具 lmstudio_vision / lmstudio_clipboard / lmstudio_models + 设置页「LM Studio 视觉识别」面板；配置持久化到 ~/.dsh/dsh-lmstudio-vision.json。Hot-pluggable — 挂载方式同 dsh-ssh：profile node_modules link + dsh.profile.bundles，无需修改 dsh 源码。

macOS desktop-control capability seam (ctx.macosUse): include-only foreground-window screenshots, exact-window observations, guarded native GUI actions, and exact app launch

Marvel hero pet runtime for the dsh web GUI: sprite-sheet animation driven by model activity, hero content registry (Spider-Man first), pet/hide/summon interactions, hot-pluggable via a profile bundle

Marvel hero skin runtime for the dsh web GUI: scoped themes, hero content registry (Spider-Man first), mouse-follow fluid canvas and identity reveal

Durable development workflow orchestration for mattpocock/skills inside DeepSeek Harness

MCP Apps Host: connects to an MCP Apps server, preserves _meta.ui in tool results, renders interactive HTML cards via sandboxed iframe, and bridges postMessage to MCP calls

MCP client bridge: connects to MCP servers and registers their tools on ctx.tools

Runtime MCP connection registry and Remote gateway: list, connect, and disconnect MCP servers from the Web UI

Memory provider registry for self-evolution

Local layered memory provider

Host-plane memory service for DeepSeek Harness backed by a local OpenViking server: automatic session capture (write-behind with retry + dead-letter), session-scoped recall, and profile injection primitives.

Independent local DSH plugin for request and response model metadata; not an official DeepSeek plugin

Zhipu AI BigModel and GLM Coding Plan LLM Provider model catalog plugin for DeepSeek Harness

xAI and Grok Coding Plan LLM Provider model catalog plugin for DeepSeek Harness

News-report runtime guard: enforces source routing (minimax for domestic, tavily for international), 24h time window, 5W1H field completeness, and archive writes for news-report skill outputs

DSH Pet Corner: a floating pet, keyless pet-image proxy, favorites, and plugin-owned settings API

Canvas preview plugin for DeepSeek Harness (DSH): a 画布 (Canvas) tab next to chat/trajectory for interactive HTML design prototyping, with canvas_preview model tool (render/annotate/clear), local file/URL preview, privacy masking and sandboxed iframe rendering.

Deployment-level Docker plugin for DeepSeek Harness (DSH): a 容器 (Containers) tab next to chat/trajectory with observe/intervene two-mode UI, 18 model tools (docker_info/ps/images/pull/run/start/stop/restart/rm/rmi/logs/exec/inspect/stats/network/volume/build/cli), read-only live shell watching, and /dock-api/* HTTP routes served via webServer.

k-identified plugin manager for the DeepSeek Harness: host-side service to list, enable, disable, add, and remove official and custom plugins

DeepSeek Harness 会话终端观察插件：在 Web 会话视图环新增「终端」页签，只读实时展示当前会话 bash 命令、状态、输出与耗时，不提供执行/交互能力。

Native DeepSeek Harness role plugin for Shiori characters

Popper: a falsification-driven correction loop for agent sessions — claim commitment, gate falsification, and mutually exclusive hypothesis revision, with an append-only evidence ledger