Plugins

Zotero library tools for DeepSeek Harness over the localhost local API (read) and the connector saveItems endpoint (write). No cloud, no API key.

Creates local, content-addressed workspace checkpoints with diff previews and two-phase restore confirmation.

Lets the model trigger compaction itself at a safe point via a compact_now tool (works mid-turn), plus a context_status tool reporting real token usage against the session's actual context window.

这是dsh的插件，语音发送弹幕。在玩游戏时通过语音输入在b站发弹幕，不切出游戏可以正常操作

Connects the Xiaozhi voice assistant to DSH Web over MCP: DSH is the tool provider, weaving 35 DSH Web endpoints into 16 voice-friendly tools for workspaces, sessions, chat, models, settings and files — outbound WebSocket to the Xiaozhi MCP access point by default (no public IP or port forwarding needed), several devices bound at once, plus a DSH settings page with live status.

Jev (TypeSafe System One) gives the human a second opinion, not the agent. When DSH's model asks you a multiple-choice question, dsh-jev-advisor builds a structured Jev request and floats the recommendation, probability spread and confidence beside the op

Local TXT reader in a right-sidebar tab plus a spoiler-safe AI reading companion: the model only sees what you have already read — the current chapter, the tail of the previous chapter, and a per-book background file it grows as you go. Excerpts, thoughts and the companion replies become structured Markdown notes exportable to Obsidian; raw text unlocks only for a book you mark as finished.

Adds standard Unicode emoji to dsh replies at the point that best matches the content, and keeps emoji out of any sentence that already carries a kaomoji.

Pins workspaces holding a running session to the top of the sidebar session list, and draws the running session row with an accent frame and tinted title, plus a short settle animation when a session stops running. Colours come from the active theme's own tokens, with the running accent taken from the token the status dot paints ongoing with.

DeepSeek Harness (dsh) plugin: live local LLM endpoint health in the web rightbar. llama.cpp slots (busy/TTFT/wedged, /metrics) + Ollama MODELS (loaded vs empty). Engine fingerprinted by response shape. Installable dsh.bundle.

DeepSeek Harness (dsh) plugin that connects Sber GigaChat models: a built-in local OAuth proxy (client-credentials -> OpenAI-compatible endpoint, upstream serialization, legacy tool-calling translation) plus automatic provider-route bootstrap, so GigaChat

Declares the OpenCode Go and Command Code route wire protocol and endpoint (so the API address never has to be typed), adds the DeepSeek V4.1 models to the OpenCode Go route and puts them first, completes the Command Code model list from its public catalog, and attaches a per-conversation x-opencode-session header to OpenCode requests (fixes 400 MissingSessionID).

Animated whale particle background for the DeepSeek Harness (dsh) Web UI - assembling from its scattered shell, breathing at idle, and being pushed around by the pointer

Routes eight DSH lifecycle events through dedup, quiet hours and digest batching into seven channels. Severity maps to the fields Bark, ntfy, Telegram and webhooks actually support; failed deliveries retry with jittered backoff and survive a restart; per-channel circuit breakers stop hammering a dead endpoint; and a self-check reports a degraded relay instead of going silent.

Speaker button beside the composer microphone — one click scrolls to the start of your newest question, marks it with a blinking caret and reads from there through the newest agent reply (dsh-tts, browser voice as fallback); press-and-drag-right picks any other reading start position on the page, press-and-drag-up opens a vertical mixer for in-page media volume and system output volume.

A `jev` tool that asks Jev (TypeSafe System One) typed yes/no, choice and score questions about one block of text and returns probabilities and confidence, with the provider and API key configured per user in Settings, a judgment ledger with its spend, and daily call and token caps.

Tiered compaction hub: replaces the single 80% compaction cliff with a configurable ladder (default 30/50/70/90) whose tiers escalate from gentle to maximum consolidation laws, folds a deterministic keyword index into every checkpoint so facts accumulate across cycles, and summarizes through a model chain — primary plus up to 3 fallbacks, each with its own reasoning effort. Adds a Compaction module sub-tab and a Settings card; toggling off always restores stock compaction.

Reopens the conversation you were last in after restarting DeepSeek Harness, by keeping a pointer file under the DSH home directory instead of patching engine files.

Speaks a short status line when DSH changes state — turn done, turn error, job done, job failed, goal complete, goal blocked, an approval waiting, or the agent needs your answer — using eight pre-recorded clips in Chinese (default) and English. The audio is played by the host process through ffplay, or through the Windows PowerShell player that ships with the OS, so it does not depend on a Web Audio tab. Windows only.

Webhook notifications for dsh sessions: a POST when an approval is requested or a turn ends, to ntfy, Bark, a Slack/DingTalk/WeCom bot URL, or any HTTPS endpoint.

Multi-instance ComfyUI orchestration: probes every configured ComfyUI endpoint for its GPU, free VRAM and queue depth, then dispatches a generation job to the idlest eligible instance. Ships three tools (comfyui_farm_status, comfyui_farm_pick, comfyui_farm_run). An instance is eligible only if it is reachable, has at least the requested free VRAM, and its queue is no deeper than the configured ceiling; eligible instances are ranked by free VRAM minus queue depth times a weight, so an idle shallow queue wins over a busy large one. Zero runtime dependencies. Optional retry on probe failure, because a single timeout across an SSH tunnel is not proof that an instance is down.

Long-term memory for DeepSeek Harness, read-only as of 0.1.0: a scribe_recall tool that reads a plain-markdown memory room, path rules that refuse directory traversal, Unicode smuggling and NTFS alternate data streams, and a bounded index whose truncation is always reported rather than silent.

Speaks your agent's reports in a voice you choose, compressing long reports before speaking. Character voices need no training - a 3-10 second reference clip clones one, and community-trained models work too - and no 6.4 GB GPT-SoVITS install: the plugin installs its own runtime and voice. An existing GPT-SoVITS can be used instead, and it is the same voice model.

Adds a speaker button immediately right of the Like button on every finalized assistant reply; click it to hear the reply through the browser speech engine, and hover it for a speed and voice panel.