Plugins

Share any excerpt of a conversation.

Connect SQLite, PostgreSQL, or MySQL databases in DeepSeek Harness: browse table structure and data, run SQL, and let the assistant query and edit data directly in the conversation.

Bounded-context memory for long-running dsh conversations and large group chats, with a focus-limited person pipeline, a fixed injection budget, and append-only event archives.

Technical English Companion helps developers learn technical English in Chinese conversations. It introduces English terms with a Chinese gloss and reuses the English name within each reply. Supports Codex and macOS DSH 0.2.1-rc.3 or 0.2.2-rc.1.

Native Windows notifications when an unawaited background job settles (completed, failed, or stopped), when a conversation turn ends, or when the agent waits for a question answer or tool approval.

Read and write your Obsidian vault from any conversation in DeepSeek Harness: daily notes, project notes, full-text search, and task check-off.

One-click restart button in the DSH conversation header: it checks for running sessions and background jobs and warns before restarting, then closes and relaunches the desktop app without opening a console window. Windows desktop app only: it restarts through a temporary Windows scheduled task, and a restart interrupts running work.

Ctrl/Cmd+F in the conversation window to find loaded messages, highlight matches, move between them, and load earlier history from the find bar.

Predicts your next instruction from the current conversation and shows it as the composer placeholder; press Tab to turn it into an editable draft.

Connect to databases in DeepSeek Harness (DSH), then query and edit data right in the conversation.

AgentBridge for DSH Desktop: run /phone in a conversation to pair a phone console with that exact live Agent.

Connect Google Drive accounts in the Settings, browse their files in a Drive tab of the right sidebar, and drag a file straight into the conversation. You create your own Google OAuth client (or service-account key) in Google Cloud Console; the plugin can read and write the Drive of the accounts you connect, and downloads a file only when you ask.

Reviews each finished turn, distills reusable skills as drafts, runs code-level checks over them, and holds each draft until the user approves or rejects it with a directive; every write is recorded in a SHA-256 ledger that supports rolling back a single entry. On by default, it reviews finished turns with the current session model (or an OpenAI-compatible endpoint set through DSH_EVOLUTION_BASE_URL), sending a bounded excerpt of the conversation, and writes drafts to ~/.dsh/skills/.staging until you promote them. Its bundle patch lowers DSH's skill-catalog description limit from 500 to 80 characters for every skill, and it starts an unauthenticated HTTP server on 0.0.0.0:18890 (DSH_EVOLUTION_PORT) that can list and promote drafts, change settings and trigger reviews.

Local-first memory governance: distills conversations into categorized memories, surfaces conflicts for human arbitration, and previews what gets injected next turn.

Long-term memory for DSH: rule gates capture durable facts from conversation turns into a single SQLite database with tiered hot, warm and cold storage; recall injects hybrid vector plus FTS5 hits into the prompt each turn; ships remember, recall, forget and cold-memory prune tools. Zero dependencies, Node only.

Per-teammate model selection for Agent Teams - a chip in the teammate's own composer pins one model for that member, or the member follows the lead conversation's current model on every step; pins survive restarts. Pins are saved in $DSH_HOME/plugin-data/team-member-model.json, and its /plugin/team-member-model routes (model catalog, setting a teammate's model) have no access check.

Running panel for dsh for skill management and visualizing skill calls: live skill and tool calls, the skill-catalog token budget against a line that follows the model's context window, a skill loadout window that enables and disables skills in bulk with dry run and undo, goals and ideas, and saved conversation excerpts. Applying a loadout moves skill folders between ~/.agents/skills and ~/.agents/skills-parked, which other apps share (dry run first, undo available); automatic translation, goal inference and idea collection call your default model only after you opt in on first open.

Layout plugin for DSH: customize the structure, order, width and visibility of the left sidebar, right sidebar, conversation area and settings page. Drag-sort and resize the three columns, per-layer teardown of sidebar/conversation/right/settings, grid slot arrangement, draw-a-frame region relocation, entry consolidation into More menus, conditional linkage rules, multiple named layout schemes, and a fullscreen layout whiteboard with a four-layer structure tree. Config persists host-side; dangerous configs have multi-layer recovery entries.

Chat with your Galatea assistant from QQ (official bot) and WeChat: the same conversation, persona and memory as on the desktop, only the paired owner's private chat is accepted, replies go back where you wrote, and more chat apps can be added. Requires the Galatea framework.

A slim tool strip beside the conversation: one-line notes into a sticky note, a follow-up or a reminder (the sticky-note and follow-up modes need the dsh-sticky-wall and dsh-followup plugins installed); selected text quoted, saved or handed to the model; saved prompts into the composer in one click; a recap of this thread or the previous one; a read-only panel for context use, tokens, subagents and open follow-ups; plus custom entries that open a path or run a command.

A conversation-independent floating window for chat.deepseek.com, rendered in a webview, draggable and resizable, with page zoom, opacity control and an optional floating ball. It makes no API calls, stores no credentials and injects no scripts.

Fully background browser automation for DSH: an offscreen browser per conversation that never steals focus or pops windows, with real CDP input.

Session install radar for DeepSeek Harness: lists what each conversation installed (pip, npm, git clones, winget-like installers) with size, path, creation time, purpose, why it was installed, and what removing it costs. Ships a sidebar panel, 10 agent tools, a timestamp sweep of temp directories, a process-command-line audit, and a two-step confirmed cleanup limited to temp directories. An optional scheduled cleanup of old temp directories is off by default.

Conversation memory behind a 20-letter recovery code: /recovery create packs the current conversation into one local record, and /recovery memory <code> injects it into a new conversation, where the model first summarises the recovered session and ends its reply with 记忆已恢复. Records are plain JSON under $DSH_HOME/recovery (mode 0600) with no network calls; the restored transcript is complete unless it exceeds the configured character budget, and /recovery where plus the dir option or DSH_RECOVERY_DIR move the directory to shared storage or a synced folder.