Plugins

CodeAct agent loop for the DeepSeek Harness: a persistent IPython session as the model's action space, with harness tools bridged in as a virtual `__dsh__.tools` module.

Data-provider MCP server for football: results, stats, odds, standings, form and head-to-head for 18 European leagues (1993-94 →) and 8 cup competitions, for any MCP client. Also a DeepSeek Harness (dsh) plugin.

Cross-agent monitoring board for local AI agent sessions (cursor-ide, claude, codex, copilot, kimi and dsh itself) with session timelines, and message injection behind a default-off two-phase confirmation gate.

Dual-model auto-routing plugin: low-cost direct / high-cost upgrade with an escape-learning closed loop (wrong direct answers auto-learn fingerprints, force-upgrading the same fingerprint next time), persisted and interoperable with the Python ModelRouter.

DeepSeek Harness plugin for NI Multisim circuit generation, simulation, optimization, and lab reports via MCP

The run_kernel_code tool for DeepSeek Harness: persistent Python and JavaScript kernels (a long-lived python3/node subprocess per session, sharing a core host driver) — a standalone plugin with no upstream harness changes required.

生物信息学「许愿式分析」插件：BioPython 全功能覆盖（任意代码执行器 bio_python + 14 个领域 skill）+ 高频语义化工具 + 零依赖自举 Python 环境（uv + venv + biopython 自动下载到插件私有目录）。用户下载安装即可许愿式分析。

Registers the remove-pdf-background-gray skill for DSH: whitens gray/off-white scan backgrounds in image-based PDFs while preserving resolution, page geometry, and anti-aliased text edges (lossless Flate write-back), via a single Python script (pypdf + Pillow + numpy).

Runs OmicVerse/OmicOS bioinformatics analyses in a persistent Python kernel, with capability-catalog search, kernel inspection and an account tab.

Make your dsh ready for serious coding. (Tools x Schemas)^REPL. `skill://`, `ctx://`, `agent://`, `dvc://`, `dsh://`, IPython REPL, Context as Variables, cross compactions recallable, full context revive. `hash-edit`, `dvc://browser`, subagent as function, workflow as function.

GEO and bulk transcriptomics workbench backed by a local Python and R engine: search and download GEO series, import your own counts or TPM matrix, then run QC, filtering, normalisation, batch correction, PCA, differential expression, GO/KEGG/GSEA enrichment, immune infiltration, survival and prognosis, signature mapping onto single-cell data, and circRNA, miRNA, lncRNA, fusion and alternative-splicing modules.

Blocks risky tool-call shapes before execution: inline multi-line Python, forwarded content without a source, missing source labels, and memory writes past the character cap. It also asks for confirmation before shell commands, and one rule is specific to the author's WeChat sending workflow.

Right-sidebar novel studio that treats a chapter as a pipeline with checkable artefacts instead of one long prompt: scaffold a project, inspect the exact context pack a chapter gets, generate 2-6 parallel candidates and keep the one you pick, then run chapters in sequence with per-chapter progress on disk so it survives an engine restart. Length band, cross-chapter Jaccard and drift ratio are enforced in code rather than by model self-grading; rejections come back as repair evidence for up to two rewrites. AI-fingerprint metrics are reported against a 3956-chapter human baseline but never block, because the distributions overlap. Ships 7 agent tools. The writing engine is a separate Python package that the plugin installs on first use. Needs Python 3.10+ on PATH (or a path set in the plugin settings).

Runs allowlisted Python hook scripts on session events and injects the text they return into the next turn, so standing rules live in editable .py files instead of the prompt. Needs Python on PATH; it only runs scripts whose names are on its built-in allowlist.

Bundles the stack-diagram architecture-diagram skill as a runtime skill, and injects a system-prompt section so the model draws diagrams without being asked. Output is self-contained HTML plus light and dark PNGs from bundled Python scripts; brand logos are fetched at run time.

Registers a skill that turns a Bilibili video link into a structured Obsidian note. It fetches subtitles only with yt-dlp and never downloads the video, merges multi-part anthologies onto one continuous timeline, and writes front matter, timestamped jump-link bullets, a summary and tags into the vault.

Translates .hwp/.hwpx (Hangul) and .docx documents into the language the session model writes - Chinese is the common case, but Korean to English or English to Korean works the same way - keeping tables, bullets, indentation and font size, and writes the result beside the original as *_zh. hwp_extract returns the document segments as JSON (50 per page by default, 200 max, next_offset for pagination); hwp_apply takes {segment id: translation} and returns the output path with applied and missing counts. The plugin never calls an API and accepts no API key; its bundled 10 MB translator only reads and writes the file locally. Windows only: .hwpx and .docx are parsed in pure Python, while legacy .hwp drives the installed Hangul over COM and requires allow_hangul.

Local Kingdee Cloud (金蝶云星空) business portal for DeepSeek Harness: open the operations dashboard in the sidebar with your own Kingdee account; data stays on your machine. Needs Python 3.10+ with Flask and requests (pip install -r requirements.txt) and a local clone of this repo linked into the profile; verified on Windows with DSH Desktop only.

Skill that transcribes local audio files into a timestamped transcript and an interactive two-column minutes page, with optional speaker separation. Transcription uses mlx-whisper in a Python 3.11 environment you set up yourself (the first run may download models); the renderer uses only the system python3. Given a public short-video share link, it can also drive a real browser through ego-browser and fetch the audio track with curl and ffmpeg, none of which ship with the package.

Multi-machine job board for DSH: 14 board tools, 4 shell tools, a GUI panel, and a dispatcher that hands queued tasks to remote hosts through dsh-remote-agent. Pick the target machine and model from the panel or the board_options tool; rename machines with display aliases while task ownership stays keyed to the host identity (machine). Failed tasks can be retried with their original goal/workspace/model, and the panel checks npm for a newer plugin version. Single-machine use starts its own loopback board; multi-machine use needs the bundled Python board server (tools/board_server.py, started with GP_BOARD_TOKEN set, binding 0.0.0.0 by default) and dsh-remote-agent hosts. The sh_* tools only reach directories on the allowedRoots list.

Runs enabled PowerShell, CMD, Bash and Python presets in a user-defined order when DSH starts, shows live output for the selected preset in a left-sidebar page, and ends each managed process tree on stop, restart, disable, delete, or exit.

Track, branch, compare and reproduce local CSV classification and regression experiments through DSH tools backed by MLflow; requires a separately installed Python backend. By default training runs as detached local worker processes that keep running after DSH exits (executionMode dsh stops them with DSH); running your own Python preprocessing or estimator code needs allowCustomCode, which is off by default; the plugin makes no network requests.

Manage iterative text annotation through DSH tools with rule versions, stale-result isolation, human review preservation and traceable exports; requires a separately installed Python backend.

Evidence-based research workspaces with archived sources, knowledge graphs, exact citations and exportable reports. Experimental alpha. Needs Node 24.9 or later, uv and Python 3.13, plus a one-time setup step (setup-engine.mjs) that copies the engine and runs a locked uv sync; pnpm must also be allowed to run the build scripts of onnxruntime-node and protobufjs.