Plugins

Four read-only tools over nine public channels — web, GitHub, V2EX, YouTube, Bilibili, Xueqiu, LinkedIn, Reddit and Twitter — with reach_search, reach_read, reach_run and reach_channels, and per-channel notes on which paths work without a login. YouTube needs yt-dlp on PATH.

Amazon product research workbench for DeepSeek Harness: category, brand and follow-seller collectors driven through the local Edge browser, a product library with unified filters, an FBA/FBM/self-operated profit calculator, blue-ocean category scoring with candidate picks, trademark lookup, supplier image search and Excel export. Proprietary software: it installs and starts without a license, but every data call returns 403 until the author issues a device-bound license key. Needs Edge started with --remote-debugging-port=9222 and signed in to Amazon; starts a local backend on port 3088. The separate 智赢 collector browser extension is optional and adds size, weight, trademark and monthly-sales fields.

Cross-session error lessons for DSH agents: search past mistakes before handling errors and reuse their verified fixes.

Three-tier memory (personal, project, session) built on the Hy-Memory engine, with relevant memories injected before each step. Ships seven memory tools (search, add, list, get, update, delete, forget), a settings panel with per-tier lists, editing, change history and duplicate cleanup, real token accounting per write, session summaries written by the engine at session end, and a user profile refreshed from recent memories. Memories can be exported to and imported from MemHub. Needs Python 3.11+ once to install the engine, or uv, which fetches 3.12 itself; uninstalling keeps the stored memories.

Full-text search over DSH conversation history: decompresses the persisted per-session session.v4.jsonl.zstd logs, matches user/assistant message text, and opens a hit at the matching message (archived sessions are unarchived first). Also binds Ctrl+F.

Tiered cold storage for DSH sessions: idle sessions move out of the live root into an on-disk cold tier instead of being deleted, an FTS5 index over that tier keeps them searchable, and a restore path moves a session back into the live root. Dry-run by default: nothing moves until you turn it off in Settings.

Full-text search over every DeepSeek Harness session, with a sidebar panel and a record_search tool.

A zero-config composer model picker with one provider per tab, independent scrolling per provider, cross-provider search with provider badges and reasoning-effort chips; selections stay in sync with the built-in /model dialog.

A searchable library of 60 Chinese task templates with variables, examples, sources and Skill methods, plus draft-safe prompt optimization using a configured DSH model.

A sidebar plugin wall with local automatic grouping, drag-and-drop organization, search, and configurable compact footer components. Replaces the built-in sidebar.

Replaces the composer model menu with a recent-models list, a searchable all-providers flyout and a draggable reasoning-effort slider; a Manage models row at the bottom of the flyout opens the Models settings directly. It takes over the composer model slot, so install only one model-menu plugin.

Chinese titles, Chinese search, hover detail cards and drag-to-reorder for DSH skills and commands in the slash menu and plus panel.

A page for writing notes inside DeepSeek Harness. · Minimal writing — one page that typesets as you type. · AI completion — a suggestion arrives when you pause, and TAB takes it. · Immersive writing — full screen, with only the page on it. · History — every note listed, searched and edited in place, and copied out as standard Markdown. · AI board — a data assistant that answers from your notes and pins the charts it builds.

Local Markdown notebook: a three-column vault (rail, folder tree, editor), split live preview, tags, stars and pins, a todo centre that gathers every checkbox in the vault and writes ticks back to the file, full-text search, session-to-note deposit, insert a note into a conversation, save an answer as a note, links back to the originating answer, vault-directory and text-size settings, and four note_* tools.

Adds a Chat entry to the DSH sidebar for project-free chats that use DSH's own sign-in and models, limited to web search, images and files dragged into the chat. Built for the DSH desktop app 0.2.2-rc.1 and tested on macOS only; do not enable it together with Starp1ume/dsh-chat-mode.

AI news panel for the sidebar: 24-hour picks, recent stories, a top-ten trending board, the daily digest and keyword search, all read from the public AIHOT JSON API. Picks and All filter by six categories; feed cards show category, source, time, heat score, summary and, when AIHOT supplies one, its recommendation note. The Picks, All and Search feeds group stories by Beijing-time day and page ten at a time with cursor endless scroll plus a Load more fallback. Read-only and client-side only, with no API key, no build step and colors taken from the host theme tokens.

Adds a left-sidebar panel to the DSH Web GUI: one ranked board of the films and series released in the past 21 days or due within the next 31 days, taken from JustWatch for the US, five European countries, Japan and Korea and from Douban for China. Each row carries a poster (or a drawn placeholder when the poster CDN will not load), a synopsis of five sentences or fewer, Douban and Rotten Tomatoes ratings and popular Douban comments. A search box filters the board as you type and asks Douban, TVMaze and the iTunes storefront about a title, merging the three into one record and listing the matches already on the board separately. Synopses, comments and titles are translated through the host translator service, so the Chinese/English switch re-renders the board. No API key and no account.

Replaces the sidebar workspace folder picker with a name box (type a name, get a same-named folder and a new workspace), adds a Move to workspace action that clones a session into another workspace and archives the original, and optional per-workspace memory that opens a continuation session after compaction and adds a workspace_memory search tool.

WSL World for DeepSeek Harness: the whole session runs in a local WSL distribution with Linux paths, clean PATH, native @/search/diffs/worktrees, plus a connection panel and project doctor. Not a wsl_* tool set.

Routes the built-in web_search tool to a self-hosted SearXNG instance and web_fetch to a self-hosted Firecrawl instance as DSH web providers, with a settings card for both backends.

She keeps the questions your chats left unanswered, then goes and looks them up later. A line she asked you that ended in "?" and never got an answer is stored as a hanging question; on a schedule inside a daily window (08:00–23:00, five a day by default; an optional mouse/keyboard-idle gate is off by default) she picks the oldest one, searches the web and Bilibili, reads what is actually readable, distils it into two or three sentences and leaves a note on her desk. Questions about its own plugins are skipped and left for you. Read-only on her events: nothing is written to her memory, her growth is untouched, and dsh-fish needs no patching. Requires the @ssamurias/dsh-fish persona plugin for her event stream and a local Ollama model; a search API key is optional.

Aggregated MCP market for DeepSeek Harness: 26 curated presets plus the official MCP Registry and user-added JSON sources, with search, source filters, connectivity tests and one-click install into the profile.

Appends one short user message before the next step when the model is circling: the same absolute path written three times, the same normalised tool error returned twice, or two steps since the last write with neither a verify nor a search. Verification is recognised from four tool names (`ops_bash`, `lsp_diagnostics`, `verify`, `run_tests`), searching from eight. Each signal fires once per turn and only after a new write or error, is silenced for the whole turn when you say not to search, and is skipped in subagent sessions. Ships the `stuck-gate` skill at bundled rank 600, so a user-level copy at rank 400 overrides its wording without touching code, plus a read-only gain panel at `/better-debug` that counts firings, sessions covered and whether the next tool call changed class; it does not estimate tokens saved.

Evidence-gated acceptance for research reports: claims, citations, data scope and method limits each need a tick plus written evidence before a deliverable can pass, with same-artifact revision history and a standalone SHA-256 receipt that says when the bytes moved.