Plugins

Per-workspace closet for temporary scripts and their run logs, kept in a .dsh-fix directory inside each project. Registers a script with its language and purpose, scans a project for leftover scratch scripts, backups and log files, runs registered scripts with a timeout and writes each run to a timestamped log, and sweeps chosen files into a reversible trash with per-batch restore. Scanning only lists and never deletes; cleanup moves files instead of deleting them; files already tracked by git are refused unless forced. Ships 7 agent tools and a settings panel.

Deliverables registry for DeepSeek Harness: agents register the final artifacts of a task (reports, pages, documents, sheets, images, datasets) into one settings section and manager page that store paths only, never copies, and remove recoverably; a standalone CLI (add, list, stats, check, remove, restore) reads and audits the same registry outside the app.

Pins a Chinese (zh-CN) thinking and answering constraint into the system prompt, and registers a per-step runtime reminder so the agent does not drift back to English after tool output. Zero runtime dependencies, single file; falls back to a second registration path and only logs a warning when the host API changes.

Composes tools, skills and MCP servers into switchable per-session sets, shrinking the resident tool surface and the system prompt on demand.

War Thunder user-skin studio: 4-point UV quad picker, quad-clipped art layers, and PNG/DDS/.blk export.

Requires Claude Code installed and signed in on this machine. Delegates coding tasks to the local Claude Code agent through the Claude Agent SDK, reports which files changed, supports one-command revert, and adds a Claude Mode toggle to the composer.

Run Python on the local NVIDIA CUDA GPU from DeepSeek Harness: device status, foreground and detached jobs, durable logs, and a measured peak-VRAM figure per run.

Toggles the rea-mcp server on and off from the DSH Web UI session header or through the rea_mcp_toggle agent tool, so the 170 mcp__rea__* tool schemas stop being sent with every request while REA is off (the tools payload drops from about 722 KB to about 95 KB) and the REA child process exits; the choice is kept in a profile-scoped state file and restored at startup by a row-level disabled expression.

Diverts operation-script writes out of the workspace into a per-session scratch root and collects each one into an inspectable vault outside the workspace, so the workspace holds only real deliverables.

Fork of fourzkw/dsh-drawai that keeps the editable .drawio canvas and its two agent tools, and adds a second vectorization engine beside the original tracer: hierarchical color clustering with Bezier spline fitting and per-channel color precision, selectable per drawing task in three fidelity levels (measured about 18, 38 and 46 dB PSNR on the same 1920x1080 frame (the upstream engine scores about 26 dB on that frame)), with worker parallelism raised to half the logical cores instead of a cap of four. The original tracer stays the default and remains the right choice for whole videos, where the new engine produces files hundreds of times larger.

Config-only DeepSeek Harness bundle that mounts the in-box MCP client bridge against CursorTouch/Windows-MCP, excluding PowerShell, FileSystem, Registry, Process and Scrape by default (16 of the 21 tools upstream 0.8.8 publishes). Requires uv on PATH.

Runs a Claude Code project's .claude/ slash commands and permissions.deny rules inside DeepSeek Harness. Command bodies expand $ARGUMENTS, positional arguments and @file references into the agent's next turn; matching tool calls are denied before they run. Every host service is probed, so a host without a command registry, working-directory service or TUI status line degrades one feature instead of failing the boot.

Requires Python 3.12+; Xiaohongshu reading needs the xiaohongshu-mcp backend and a QR sign-in (ffmpeg/OCR optional). Batch-read Xiaohongshu notes and WeChat public-account articles: per-note body, image text via local OCR, video frame text and comment samples, structured JSON/Markdown output, and the original source URL on every result. Read-only: it never publishes, likes, follows or comments.

Reference search for DeepSeek Harness: refkit_search looks up image, video, audio and text references across 24 open sources and tags each result with its normalized license and a use verdict for the intended use; refkit_rights re-checks one license against another use. Source keys and search tuning live on a settings page. 11 of the 24 sources need API keys; the use verdict is returned only when an intended use is passed. Runs on DSH ^0.1.7-rc.2 || ^0.2.1-rc.2 (tested on 0.1.7-rc.2 and 0.2.1-rc.2).

Motion graphics from a dsh session through the Miran Fann engine over MCP, with a Miran settings page and motion-design skills. Requires the Miran Fann engine installed separately (npx miran-fann setup) and a signed-in install.

Run the local Cursor CLI agent on a coding task in a workspace and continue the run by chat id.

A philosopher's way of thinking made executable: dialectics and practice-testing to govern hallucination and long-session drift, so done is a verifiable fact rather than a claim.

Read Bilibili and Douyin from a session: WBI-signed Bilibili API calls behind one shared pacer (4 req/s, per-host cooldown after a 412) for video, feed, search, author, comments, danmaku and subtitles, plus Douyin hot list, video, feed and search rendered in a locally launched Edge over CDP.

System One decision runtime for dsh. Ships fast non-generative tools (choice, score, screen, gate) on a local or hosted System One gateway, and — without the model calling anything — prunes oversized tool output segment by segment, refuses repeated identical calls once System One confirms stagnation, verifies completion claims against real tool evidence, and corrects failing retries. Published on npm as dsh-harness-reflex (plugin) and harness-reflex (core, also a stdio MCP server).

Algorithm field guide for graphics, animation, physics and AI work — 219 entries with tradeoffs and links to existing implementations, exposed as 10 agent tools and a sidebar page. AI-assisted tools require DEEPSEEK_API_KEY.

Per-workspace gate that hides the tool schemas and system prompt sections of selected plugins from the model in directories that do not need them.

PCL-style modpack for DSH plugins: export a small .dshpack.zip manifest of the selected third-party plugins and their versions, then import it with a preview before reinstalling.

Model-driven tool-result archiving for DeepSeek Harness. Results above a configurable size carry a short inline id (\[#a1b2c3d4]); the model archives the ones it no longer needs into the session's own archives/ folder and the context keeps a small recoverable stub with the file path. Originals stay in the append-only session log, so nothing is lost. Ships a settings panel with per-size-bucket recovery stats, thresholds that apply without restart, and a conditional reminder when results are waiting to be archived. Requires dsh-better-sidebar for the statistics panel.

Genspark tools for DeepSeek Harness - web search, crawl, image/video/audio generation, media analysis, X/Instagram/Reddit search and model listing through @genspark/cli, with 17 typed tools and 18 /genspark-* commands. Requires @genspark/cli installed and logged in (gsk login), or GSK_API_KEY set.