Plugins

Repackages the MedXpert skill collection (zhaoxinghua09-cell/medxpert-skills, Apache-2.0): twenty-one medical-device regulatory skills for NMPA, FDA, EU MDR, and PMDA workflows. Do not install the same skills twice (do not also copy them into ~/.dsh/skills).

Development-preview A2UI v0.9/v0.9.1 conversation surfaces using official schemas and renderer, session-scoped form actions and non-executing history replay; verified only in bounded synthetic DSH 0.2.0-rc.2 workflows.

Four-tier prompt enhancer for the composer — the system tier always emits a LangGPT-structured system prompt (# Role / Profile / Skills / Rules / Workflows / Initialization), upgrading task drafts into the system prompt of the agent that runs them, while the other tiers rewrite as a light tidy-up, a single paragraph, or a five-section brief; the iterate workbench refines an editable base under an optional requirement, with fill-back and undo. Never changes intent, never invents requirements.

Graph-based code context, impact analysis, review workflows, interactive graphs, and permission-controlled refactoring for DSH using the code-review-graph engine.

Adversarial review for DeepSeek Harness: put one settled plan on the stand, send read-only red-team experts at it by dimension, then split their findings into single defects you endorse or reject one by one. It never rewrites the plan and never fixes code.

Bridges LobsterAI (Youdao Lobster) into DSH: its free daily check-in credits become usable inside DeepSeek Harness, so conversations, workflows and skill packs run on those credits. Requires the LobsterAI desktop app (Windows); new-account login and gateway sync need the separate LobsterDaddy console (not in this repo). Unmaintained since 2026-10-21.

Conversation-level multi-agent workbench for editable task planning, real child-session observation, plan-versus-runtime comparison, and evidence-backed recovery previews; tested with DSH 0.1.1-rc.2.

SOP-driven business Agent Teams for DeepSeek Harness: configurable captain/member agents, human- or agent-defined SOP workflows (nodes, dependencies, quality gates, manual gates), event-driven task scheduling, and a web dashboard (SOP canvas, run board, a

Triggers GitHub Actions workflows and local test pipelines, streams their logs back, and on failure hands the log tail to the agent for diagnosis.

Local, auditable personal memory for DeepSeek Harness: preferences, project conventions, workflows and error lessons persisted as a changelogged JSON file in the workspace, with onboarding, auto project-convention extraction and evidence-backed lessons.

Connects DSH to local Beav for Xiaohongshu (RED/RedNote) and social-media AI operations: topic research, knowledge, copywriting, image, audio, and short-video workflows with approval and artifact tracking.

AI-assisted research workbench with editable step workflows, on-demand skill binding, a live execution graph and in-project document review (needs a local Python with python-docx, python-pptx, openpyxl and PyMuPDF), while keeping existing projects on their original configurations. Requires DSH 0.2.2 or later.

Persistent concurrent node-graph workflows for DeepSeek Harness with agent, command, HTTP, human, fan-out/join, gate, and subworkflow nodes.

Runs the agent through GEML workflows: work items move through flow steps, practices are recorded at the step they belong to, and a board tab in the right sidebar shows where each item stands. Also mounts the GEML MCP server and skills; when the project has a fixed workflow, a supervisor gates tool calls by its current state. Needs Node.js 22+; the GEML MCP server is fetched with npx (@geml/geml) when a session starts.

Reusable workflows: orchestration scripts saved as files under .dsh/workflows, discovered from the project, and run by name with arguments, so they live in git and can be shared. Four tools: list, show, run, save.

Privacy guard for git workflows: before git commit, git push or gh repo sync it scans the content that would enter git (local paths, user info, API keys, private keys, connection strings) and blocks anything not approved in the DSH approval UI; approved items go into a manageable, masked whitelist. Native DSH Config schema with live settings, first-push full-repository scan, fail-closed checks, zero npm runtime dependencies.

A local ComfyUI workbench for DeepSeek Harness (DSH): image/audio/video generation, W2/W3 sprite cards with Qwen back-view walk-up, local Flux.2 Item Studio assets with guided SDXL fallback, Hugging Face downloads, inspection and arbitrary workflows.

DIY workflow builder for DSH - compose multi-step prompt workflows with self-check gating, engine-enforced jump rules, agent-driven execution and a live run panel.

A board for DeepSeek Harness that composes stations, workflows and projects from separate Git repositories; GitHub is the access gate.

Leanary: a mathematical research plugin for DeepSeek Harness, with persistent research workflows and Lean verification.

Embedded identity and payment-governance trust plane for multi-agent cross-border payment workflows.

OpenSpec spec-driven workflows for DeepSeek Harness, generated from OpenSpec 1.14.3: twelve /opsx-* slash commands that carry the workflow prompt plus the arguments you type, and twelve openspec-* skills the model loads on its own.

Connects a ComfyUI server to the agent through five draw_* tools (status, workflows, models, generate, reverse) and a sidebar panel. It ships no workflows and assumes nothing about your models: every API-format JSON you drop into the workflow directory is inspected to find where prompt, negative, width, height, steps, cfg, seed and LoRA live, and a sidecar file can override what was detected. LoRA injection splices into any graph. All paths are configuration, editable from the panel and saved to a settings file, so it works against a remote or already-running ComfyUI. Headless auto-start and idle shutdown are opt-in and off by default.

Production pack for ComfyUI inside DeepSeek Harness: submit API-format workflows to a local or remote server, batch-generate and fetch outputs, and diagnose bad images with measured parameter ranges and a failure-mode decision tree (black frames, duplicated subjects, color fringes, OOM).