Plugins

DeepSeek Harness 插件：把一段自定义 Markdown 提示词注入每次模型请求（独立上下文注入行），可在设置页编辑，并支持从 Claude Code / Codex 一键导入规则。

Adds an "approve for me" access mode whose approval requests are answered by an independent reviewer model instead of a person: the reviewer can read the workspace read-only, a reviewer that cannot run hands the request back to the human, and every decision is recorded with its rationale in an Approvals tab.

DSH host plugin: promote the workspace you just submitted a task in to the top of the sidebar — once per submission, and never while the task is running. 提交任务时把该工作区置顶，任务执行期间不重排。

Deny an AI agent access to .env files, credential stores, keys and any path you hide — across file tools, shell commands, search selectors and run_code. A DeepSeek Harness (dsh) plugin.

开口即成文 · Speak-to-prompt for DeepSeek Harness：云端 ASR 语音识别 + 提示词优化 + 填入草稿/自动发送，跨平台 macOS / Windows。

Report stage — academic paper writing for asset pricing and machine learning research. Ships one skill.

Data stage — data discovery and quality assessment for asset pricing and machine learning research. Ships one skill.

Literature review stage — search, synthesis, and citation-chain tracking for asset pricing and machine learning research. Ships one skill.

In-browser files pane for the DSH Web GUI: a Session View tab that browses the host's files and reads them in place

Tavily-backed search provider for dsh: registers the `tavily` provider on ctx.web so the web_search tool runs against the Tavily Search API

A dsh plugin that replaces the built-in llm-pi-ai and llm-deepseek adapters, the model selector and the Models settings page with self-maintained versions, tracks upstream @earendil-works/pi-ai, and adds quota/balance lookups plus a provider management UI

Model training stage — end-to-end analysis, model training, and estimation for asset pricing and machine learning research. Ships one skill.

Sets the DeepSeek flash model's reasoning effort per step — thinking off for short cheap prompts, low for a plain request, high for engineering work — and raises it only on repeated tool failures. The model itself never changes; effort max is opt-in.

DeepSeek Harness Web GUI plugin: real-time spend for the current session, priced from per-model token rates and shown as a pill under the composer

PPT stage — presentation generation (Beamer/Quarto) from the paper for asset pricing and machine learning research. Ships one skill.

DeepSeek Harness plugin: customize the Web GUI fonts (interface family, code family, and three font-size axes) from Settings

Feishu (Lark) OpenAPI MCP connection for DeepSeek Harness: drives the official @larksuiteoapi/lark-mcp stdio server with your Feishu app credentials, exposes Feishu API tools (IM messages, Bitable, docs, calendar, drive) under mcp__feishu__*, and a web se

Automatic tier-based model routing for DeepSeek Harness: a virtual `smart` model classifies each request by difficulty (hard / normal / easy) and by vision need, then delegates it to the models you already configured. 三级难度 + 视觉自动路由，虚拟 smart 模型零配置接入。

Composer add-assets plate for the DeepSeek Harness Web Client: an options menu on the composer's + button (files, folders, slash commands) and a Claude Code-style draft attachment preview

DSH web plugin: a VS Code-style code editor view (file tree, Monaco, ripgrep search/replace, Markdown preview, git diff).

DeepSeek Harness plugin: discovers the npm/pnpm/yarn/bun scripts of the current workspace (never descending into node_modules) and runs any of them in the background from a compact control in the composer area or the session header.

Community DeepSeek Harness plugin: deterministic engineering orchestration with Commander, Executor and Smart Watchdog (CxService + cx_controller tool).

One-click restart for DeepSeek Harness: a web button (plus a dsh_restart agent tool) hands the relaunch to a detached helper that waits for the port to free, relaunches the exact same command, streams the new host's output, auto-reconnects the page — and

Community ACP bridge for DeepSeek Harness that supports session/load by replaying a persisted session's transcript