Plugin
Sfoglia, filtra e installa i plugin di DeepSeek-Harness.
57 plugin trovati
dsh-liquid-glass
shuoweifu/dsh-liquid-glass
A high-performance WebGL liquid glass plugin made for DeepSeek Harness. It features domain distortion, chromatic refraction, and a 5-layer GPU overheat protection system.
ai-voice
root-7-khx/ai-voice
A local AI voice studio for DeepSeek Harness with nine local TTS engines for voice design, voice cloning, emotional performance, a reference-audio library and audio export. It only registers /ai-voice/api routes; the UI lives in the companion-memory plugin's integrations page, and every engine needs its own Python virtualenv and downloaded model weights, which are not bundled.
dsh-session-vacuum
mili1360/dsh-session-vacuum
Deletes the write-only streaming fragment rows from session logs on disk so a long session loads into fewer objects; the pruned log is checked against the loader contract before the file is replaced, and the previous file is kept as .orig.
dsh-lag-guard
456erddgr/dsh-lag-guard
Limita a 120 messaggi la richiesta della cronologia e sostituisce con un array vuoto gli schemi request/header oltre 200.000 caratteri prima del rendering React, così le sessioni lunghe si aprono senza blocchi.
dsh-performance
wanyi424/dsh-performance
Leggero e snello? Mostra CPU, memoria e blocchi UI DSH suddivisi per plugin, terminale e server MCP, li confronta con gli orari dei blocchi e consente di copiare un rapporto diagnostico. Solo Windows x64.
dsh-lucky-block
lyhtakeashot/dsh-lucky-block
Aggiunge un pulsante lucky-block trascinabile nella web UI DeepSeek Harness. Clic: brevi effetti gratuiti (popup, finto blue screen, fortuna del giorno, trivia, battute fredde, musica, stunt visivi). Pressione 1,4 s mina il blocco e apre una sessione per paper casuale, news, invenzione o racconto breve. Aggiunge una scelta lucky block alle domande multiple choice del modello e sorteggia; «rollback» annulla. Aspetto/pesi regolabili.
dsh-speed-gauge
moreanp2003/dsh-speed-gauge
Gauge in tempo reale della velocità di generazione DSH: quadrante semicircolare e curva trend fluida, con 50 campioni tok/s al secondo per misurare con maggiore precisione.
dsh-cpp-optimization
chthollyfan/dsh-cpp-optimization
Skill di audit C++: rileva lo standard C++ effettivamente usato dal progetto prima di formulare raccomandazioni. Controlla RAII e ownership, semantica di move, sicurezza delle eccezioni, prestazioni e memoria, container e algoritmi, concorrenza, template e compilazione, limiti dello standard e type safety, ABI e interfacce. Ogni rilievo include punto di controllo, motivo, esempio da errato a corretto e standard minimo.
wincu
dxxcao/wincu
Controllo del computer Windows per DeepSeek Harness: backend PowerShell UIA persistente tramite 16 voci MCP aggregate (read/click/type/observe/verify). Policy rafforzata (limiti inderogabili, approvazioni, collegamento all’osservazione, controlli su proprietà del punto e focus) e ottimizzazioni (prewarm, cache, verifica della stabilità con hash dei pixel). Fork di dsh-computer-use-win.
dsh-wsl-perf
173787247/dsh-wsl-perf
Prestazioni host Windows da WSL: CPU, memoria, disco e processi principali.
dsh-pianist
laplace-bit/dsh-pianist
Plugin per pianoforte: chiedi all’agente di suonare un brano e questo viene riprodotto su un pianoforte a coda Canvas2D con campioni reali Salamander Grand, un palco immersivo e una tastiera interattiva di 88 tasti.
dsh-k6
wode25500/dsh-k6
Test di carico con Grafana k6: prima smoke-test, poi esegue gli script e legge i percentili dal riepilogo JSON.
dsh-compressor
lifeodyssey/dsh-compressor
Un port snello di Headroom: comprime l'output dei tool, taglia fino al 20% del contesto, senza intaccare la cache del contesto del modello o le prestazioni dell'agente.
dsh-plugin-code-review
yytbit/dsh-plugin-code-review
Skill di revisione strutturata del codice per DeepSeek Harness -- bug, sicurezza, prestazioni, stile
dsh-theme-obsidian
inh-ui/dsh-theme-obsidian
Desktop-style client theme for DeepSeek Harness: translucent slate sidebar, #151517 main canvas and #2c2c2e composer dock, four palette presets (obsidian-origin, glacier-midnight, titanium-noir, sonoma-frost), a settings panel for blur, sidebar opacity, ambient glow and content radius, and three GPU performance tiers (ultra, balanced, eco).
dsh-tts-voice
pandaman0759/dsh-tts-voice
End-of-turn voice summary: the model calls voice_say once before finishing a reply so the conclusion is spoken through Breeze TTS, with Chinese bracket audio tags (laughter, sighs, pauses) and optional performance instructions. Audio is synthesized host-side and played from browser memory without ever touching disk; the character limit is a writing target for the model, not a cut-off, so long lines still play in full. The API key stays in the credential store, and an optional daily Breeze check-in is included. Needs a Breeze API key and sends the spoken text to the Breeze service; its HTTP routes (including the settings route that stores the key) have no DSH access check.
dsh-hr-workbench
xscanzm/dsh-hr-workbench
Runs a complete HR workbench inside DSH — a 3D office, recruiting pipeline (resume, interview, written test, score), employee roster, files, attendance, performance, training and analytics — with the site’s AI assistant answering through the model DSH already runs.
dsh-model-radar
chip-g0202/dsh-model-radar
Runs a four-task, 150-point model intelligence benchmark inside DeepSeek Harness (Qin 60, Pompeii 45, Pelican 22.5, Candy 22.5): each selected model × reasoning-effort pair executes every task in its own isolated subagent session, Qin and Pelican are graded by the selected judge model against rubrics, Pompeii and Candy programmatically, and batches keep scores, artifacts and cost. Adds a 0–1000 real-world performance score per model × effort computed from local session logs across seven dimensions (output speed, first-token latency, reasoning speed, tool cadence, tool latency, tool success rate, cache hit), with baseline deltas and controlled speed runs, plus a sidebar page with ranking, performance, history and settings views. Each batch spends real tokens of the candidate and judge models, and candidate subagents get file and bash tools inside an isolated run directory.
dsh-aot-runner (plugin)
boyingliu01/dsh-aot-runner
Drop-in replacement for the Node.js subprocess runner with a NativeAOT binary, cutting per-spawn overhead by 59-67% (p50 156ms to 63ms) while staying byte-identical on 28 A/B cases; Windows x64 only, requires a one-line upstream patch, and ships disabled until you set enabled: true.
dsh-longsession
asots/dsh-longsession
Adds a reversible “Collapse history” toggle to the session header with a light/dark status dot: it unloads the history already rendered by the page and keeps the latest two turns, then shows “History collapsed”; clicking again restores the saved window. A manual restore keeps auto-collapse off for the rest of that session; thresholds and request caps live on Settings → Long session.
dsh-agent-speedup
hj1h/dsh-agent-speedup
Tool-execution efficiency suite for DeepSeek Harness: read-only result cache, speculative prefetch, and concurrent readonly shell commands. Honest scope: helps when tools are slow and calls repeat; model latency dominates otherwise.
dsh-token-rmb
liuyun2025/dsh-token-rmb
Appends the current turn's estimated token cost in CNY to the assistant action row, right after the usage/time indicator. Prices come from a built-in DeepSeek peak/off-peak CNY table (Beijing time, Monday-Friday 9:00-12:00 and 14:00-18:00) and can be overridden from the profile's cordis.patch.yml; a model without a configured price shows a dash instead of a made-up amount. Follows the shipped Performance & usage - Compact setting and hides together with the usage pill. Token cost only, per turn only.
dsh-windows-monitor
joyboy-yy/dsh-windows-monitor
Overlay di monitoraggio sistema per DeepSeek Harness: una piccola pillola nell’angolo mostra CPU e memoria; espansa, mostra attività e velocità del disco, capacità per volume, rete, uptime e batteria. Solo Windows.
dsh-perf-hud
yanzhaohui1999/dsh-perf-hud
HUD prestazioni client-side e recorder continuo: FPS/peggior frame, attribuzione long task in 4 livelli (React fiber/subtree/DOM writes/scheduler), callback rate/tempo sync per plugin, crescita/scrittori DOM, JS heap, event loop host e RTT. night-watch conserva gli stalli peggiori ed esporta incident pack riproducibili.