Plugins

Cost tracking priced from an online catalogue (models.dev + OpenRouter, 8000+ models, each call priced at its own instant): a per-turn cost line, a Cost tab attributing spend to tool output / model output / system prompt / commands, budget, monthly forecast, ~166 live-rate currencies, and a bill_stats agent tool.

Animated Claude-style reasoning-effort slider and model picker that replaces the native DSH model selector.

DeepSeek Harness plugin that shows API balances and available models in Settings

MCP server installer and verifier for DeepSeek Harness: interactive init writes real patch entries into a profile, list and validate inspect them, and a CI-enforced verifier connects to every curated server.

Pops up a mini-game menu (wordle, match-3, extensible) while the model generates.

HUD status panel: Git status, MCP servers, skills, model and token usage in a floating side panel.

Design-review gate for DeepSeek Harness, a DSH port of Google's Elephant-Goldfish Model: a context-rich Elephant session drafts the design, then fresh, zero-context, read-only Goldfish reviewers stress-test it before implementation. Registers the `egm` plugin id; install one EGM bundle per profile.

A single-character agent that lives across QQ (via SnowLuma / OneBot 11), Bilibili and Douyin (the Douyin channel needs an always-on Windows desktop): persona cards with slowly evolving traits, per-person memory with Ebbinghaus decay and correction, a virtual clock driving a daily routine, proactive messages, auto-extracted memories, and a ten-tab settings panel covering retrieval, persona, routine, memory inspection, channels, habits, models and usage. QQ groups are deny-by-default: a group is answered only when it is on the allowlist or allGroups is explicitly switched on, and it still has to @-mention the bot. Private chats answer QQ friends only. Bilibili/Douyin auto-reply is off by default, limited to the owner, allowlist and mutual follows on Bilibili, and to the owner and allowlist on Douyin unless reply-to-everyone is explicitly enabled. Vendors MIT-licensed code from dsh-qq-bridge and dsh-douyin-bridge, credited in-source with their LICENSE files; it does not need either installed alongside.

Antu diagrams inside DeepSeek Harness: Host tools (antu_validate / antu_layout / antu_render) and an interactive chat card.

Per-directory access gate for DSH: deny / read / full levels for names and paths, roster groups, per-session overrides and an audit log, plus seven model-facing `gate_*` tools whose widening requests must be approved by the owner in the panel. Installs with `dsh plugin --profile desktop add dsh-isolation-gate`; MIT, no network calls, state stays on disk.

DeepSeek Harness's own fast mode: the same model on the same reasoning effort, fewer output tokens, all prompt-side. Only the output-discipline switch is on by default (late mode, active from step 4); every other switch is off and can be enabled on its own; turn on the tool-output trim switch when tool output is long. Measured on one machine, not a leaderboard: 30 task pairs with long output at the trim switch 0.756x wall clock and 21% fewer output tokens; 24 pairs on 6 real pallets/click PRs 0.805x and 23% fewer. On official SWE-bench Lite instance definitions with a self-built harness it does not show a gain (8 instances, three independent runs pooled to 72 pairs: 0.933x wall, 0.903x tokens, wall still crosses 1; single runs of the same configuration differ by 30%), and stacking switches on top of the trim switch is worse than the trim switch alone. Numbers and repro in the repo.

Intelligent UI for DeepSeek Harness: the model answers with an openui-lang program and the Web GUI renders it as a live, interactive UI inside the chat

Codex-style dark interface for DSH on Windows, with navigation, project and conversation sidebars, and model and reasoning controls.

Brings the WorkBuddy, Trae and Qoder desktop logins on this machine into DeepSeek Harness as OpenAI-compatible local endpoints, with an account pool, automatic account switching when WorkBuddy is rate limited, and a settings panel.

Bring the WorkBuddy accounts signed in on this machine into DSH, managing CN and international accounts separately. Supports automatic failover between accounts, one-click refresh of model catalogs and credits, the CN daily check-in, and a togglable OpenAI-compatible endpoint.

Connects one existing DSH session to QQ through OneBot/NapCat without starting a new session, and uses a fail-closed whitelist plus @-mention, nickname and configurable-probability wake rules to decide which messages reach the model.

Multi-host SSH workbench: keep remote servers in a host library, grant them per project, and open an interactive terminal or dual-pane SFTP in the right sidebar, while the SSHManager tool runs remote commands and transfers files without exposing keys to the model.

Per-message delete for DSH: remove a user message, one reply step, or a whole assistant reply from the derived model context through the official surface-replace contract, and hide it from the visible transcript without rewriting the append-only log.

Serves the live opencode Go model catalog from the harness's built-in opencode-go provider route.

Routes model requests, web fetches/searches and spawned tools through the Windows system proxy or a custom address — idempotent kernel (reads never rebuild the dispatcher), header one-click toggle, per-URL route diagnosis and NTLM-aware 407 hints.

ClineBot and ClinePass integration for DeepSeek Harness: dedicated settings page, live quota dashboard, in-UI key saving, curated and custom model selection, and a /cline command.

Adds ChatGPT-style stop, edit-and-resend, native quote annotations, and an independent sidebar follow-up chat with model and image input for DeepSeek Harness Desktop.

Pins every request to the native max-effort tier and adds a per-model/per-mode reasoning-depth panel with a 0-20 capability slider, isolated Velocity/Apex profiles, and offline reasoning toolboxes. Distributed as an obfuscated build; commercial use allowed, reverse engineering closed.

For the DeepSeek Harness native vision model deepseek-v4-flash-vision-exp, raises image admission limits to 32 MiB / 8192 px / 600 images and adds a highres_read tool that tiles large images, then returns the whole image plus 800x800 tiles through the host read_image tool.