Plugins

Cleans the harness prompt-authority layer into a reviewable patch set, writes prompt-inject.md and lets each model use a different default prompt, reports which patches took effect and whether the running host loads app.asar or an unpacked tree, and swaps a sealed install to the patched tree once the desktop process is idle. Includes a status page and the dsh-purge CLI.

Data quality checking for DeepSeek Harness — profiling, cleaning, and verification pipelines with structured reports.

Read, create and edit Figma designs through natural language in DSH, including batch layer edits, components, variables, screenshots and asset exports. Local mode does not use the official Figma MCP call quota.

Puts the Markdown documents, source files, images and video that already exist anywhere in the session workspace into the DSH sidebar, ranked by relevance to the current conversation. The default lexical mode matches your recent messages against title, summary and body locally with IDF weighting (BM25) — no model calls, no network, nothing leaves the machine. The list comes from scanning the workspace rather than from a recent-files list, so restarting DSH or opening a new session does not empty it. Results are organised into a three-tier reading order (primary / supporting / related) with a verifiable reason on every row — which field matched, which document cites this one — and one click opens an in-place preview: Markdown as rendered text, relative-path images actually drawn, source code with the host's own syntax highlighting, video streamed over HTTP Range. The same Context Pack is exposed to the agent as a knit_docs tool. Later versions added read evidence and lifecycle tracking (attribution is frozen at the moment of reading), pin / exclude (exclusion applies before retrieval, and is undoable), relation evidence (references / imports / tests, with line numbers) and an incremental index reused across restarts. Since v0.23 there is also an optional hybrid mode: it is off by default and only runs once you configure an Ollama or OpenAI-compatible embedding provider yourself. When it runs, what leaves the machine is your query text plus each candidate file's title and summary — a Markdown summary is an excerpt of the body's first paragraph (at most 60 characters), a code file's summary is its workspace-relative directory, and an image or video contributes its file name; full bodies and absolute paths are not sent, and a loopback provider keeps everything on this machine. Since v0.24 it also associates uncommitted changes: a read-only git status (fixed argv, no shell, whitelisted child-process environment) combined with the same reference / import / test / mention evidence shows which files a change is likely to drag along, and it is off by default. The benefit of semantic retrieval still awaits a real-provider run, so no recall improvement is claimed.

Off-peak batch runner for DeepSeek Harness: users define tasks during busy hours — prompt, strategy, priority, workspace, locked files — and release them by hand; the plugin then runs the batch automatically inside user-defined off-peak windows. Each run passes five preflight checks (workspace, git HEAD, locked-file sha256, window fit, daily budget), supports four execution strategies (single / iterative / sampling / review) and three autonomy levels (L1 per-task, L2 batch, L3 full-auto), and closes the loop with a morning report containing diffs, actual spend and savings from a daily ledger aligned with official peak/valley pricing.

Use a native browser sidebar or @ to reference logged-in Chrome/Edge tabs, authorized local HTML and running Windows apps. Read compact snapshots, batch actions, capture screenshots and use available WebMCP tools; supports user-authorized file uploads, same-frame element drag/file drop and opt-in network metadata capture. Page/app context appears as expandable chips; dockable element annotations and selected local diagnostic logs append to chat drafts without sending. App control prefers a matching installed MCP or approved installation, then falls back to Cua Driver with text, hotkeys and paged controls; desktop reads present the window once per connection or when recovery is needed, and failed actions retain evidence and stop without automatic replay. Background input depends on the app. Requires Chromium-based Chrome or Edge 125+ with the companion MV3 extension loaded unpacked and paired once; installing the plugin does not install the extension.

Local-first A-share research workbench for DeepSeek Harness: market dashboards, watchlists, valuation, four investor agents, versioned reports, and continuous post-report chat.

QA project workbench for dsh: manage test projects, requirements, test cases, defects, milestones, reports and approval gates through a browser UI and 18 QA-domain tools.

DeepTrace — deterministic agent reports from session logs (daily/weekly/monthly/yearly/custom): cost & token breakdown, 8 insight rules, collaboration review, live provider balance, PDF/PNG/HTML export; read-only, never rewrites history.

AI image and video studio as a floating DSH panel, so creation runs alongside the conversation instead of replacing it: text-to-image, image-to-image and multi-image composition at 1K-4K across eight aspect ratios; text-to-video and image-to-video with first-frame control at 4-12 seconds; short-drama mode that imports a script (.txt/.md/.json), breaks it into storyboard shots for preview and batch generation; and a prompt-expert workspace. Zero runtime dependencies.

Session-bound persistent TypeScript REPL for DSH PTC mode: bindings and imports stay live across run_code calls, edit_run_code replaces one line without resending the block, module import/export syntax is adapted via AST rewriting, and durable bindings are restored from the session journal after a cold restart.

Install safety net for DeepSeek Harness: pre-install snapshots, one-click/automatic rollback, guarded boot, and incident reports that auto-trigger agent analysis. 中文: DeepSeek Harness 插件安装安全网（安装前自动快照、一键/自动回退、守护启动、事故报告自动触发 Agent 分析）。

Windows only: the agent edits documents inside the Word, Excel and PowerPoint (Microsoft Office or WPS Office) windows you have open, through the applications' automation interface, so changes appear live and you can edit alongside; it reads structure, applies batches of edits, renders pages to images to check layout, and saves or exports PDF.

Imports Claude Code setup into DeepSeek Harness from a native settings screen or CLI, with dry-run previews for skills, commands, agents, hooks, permission rules, and MCP config. Codex and OpenCode remain supported as secondary origins, with collision-safe apply and post-move checks.

Mobile-first Web UI quality-of-life pack for the DeepSeek Harness: a Chrome-style active-session tab bar, full-screen sidebar swipe open/close, overlay sidebar, collapsed-sidebar recents, keyboard-suppressing session switching, viewport/IME keyboard adaptation, touch feedback, code/table inner scrolling, hidden permission dropdown, a full-screen settings rewrite with tab memory, and a compositor-driven status pulse, no-touch-drag long-press protection — 14 features, each independently toggleable from the Settings QoL section with instant effect, persisted per browser in localStorage; pure client-side CSS/JS, mobile rules locked to viewports of 768px and below so desktop stays untouched.

Multi-platform research and deep search pipeline: harvest_scout for parallel discovery across GitHub, Web, Twitter, Reddit, Xiaohongshu, Bilibili, and YouTube, harvest_deep_research for automated comprehensive research reports, plus extract, cross-source verification, and credibility auditing.

Host-wide MCP server panel: imports the servers declared in Claude Desktop, Claude Code, Cursor and Codex CLI configs, searches six upstream catalogs and installs from whichever of them declares an install plan, asks only for the fields a server actually declares, and checks each row for whether the mount, the handshake and its tool list really work.

Pending-change review for DeepSeek Harness: every `edit`, `write` and `str_replace_editor` change is collected automatically into the file list, where a single change block, the blocks within a selection, or the whole file can be kept or reverted, and undo/redo is supported. It imports the workspace's version-control changes, shows the diff in a single-column or side-by-side view, and provides a Markdown preview in which changes can also be kept or reverted, in-file search, block jumping and copyable line references; the diff data is stored persistently, so the review continues after a restart.

Exports the session event stream to OpenTelemetry OTLP and Langfuse as sanitized, buffered traces and metrics, off by default.

Persona engine for DeepSeek Harness: one config.json drives the AI's self-name (per model, per tier), how it addresses you, relationship stance, character text, per-item work contracts with individual toggles, thinking-chain language, appearance and reply tone (both opt-in and per-model overridable), and long-term memory gated by human confirmation - the AI can only append proposed candidates, a human confirm line flips them into effect, and the log is physically append-only and replayable. Confirmed entries are injected as a quoted data-not-instructions block; entries tagged with a kind can instead be routed to arbitrary files on confirmation and never reach the prompt. Three prompt sections re-evaluate from disk on every assembly, so edits apply on the next step with no restart. Persona presets are shareable JSON snapshots (imports and exports SillyTavern character cards, v2/v3 read, v2 written) and deliberately never carry memory. Zero runtime dependencies, no network, no shell; any error degrades to an empty section rather than a broken session. Ships blank - installing changes nothing until you write something.

Encrypted local credentials vault for the Harness: a web settings page and vault_* tools to store, search and copy passwords, API keys, TOTP secrets and card data, with health audits, expiry rotation, imports/exports and read-only/ask access modes.

Local-first four-bucket Token observability: persistent session/provider/model/day dashboards, trends, budget and anomaly signals, public-rate estimates, safe aggregate exports, and explicit opt-in usage or session-trajectory analysis.

Offline schematic review for EasyEDA Pro, exposed to dsh as four tools — one-shot checkup reports, architecture skeletons, an install doctor, and a live-editor bridge that prefers read-only actions, all driven by the local boardwise CLI.

A WYSIWYG Typst editor for papers, reports, exams, resumes, charts, posters, sheet music, and slides exported as PPTX.