Plugins
Browse, filter, and install DeepSeek-Harness plugins.
396 plugins found
dsh-theme-animal-island-ui
shirolin/dsh-theme-animal-island-ui
Animal Island theme with boarding-pass masthead, cream notebook paper, 118 ThemeRuntime tokens, in-app appearance presets, and starry night mode.
dsh-local-trim
leonlee114514/dsh-local-trim
One-step local model onboarding plus a light harness for DSH. On load it probes loopback and private-network inference endpoints (llama.cpp, Ollama, vLLM, LM Studio), reads each model's context window and output budget from the backend, and registers the provider through settings, so a local model becomes selectable without hand-writing config. It then trims only that provider: masks the tool list through tools.restrict, drops named system-prompt sections and memory contexts, swaps the injected instruction block for a short one while preserving its first 80 and last 96 characters, and truncates tool-schema descriptions without touching structure or required fields. Cloud providers are returned as the same assembly object, untouched. The injected block is re-read from disk on every assembly, so editing the file takes effect on the next request with no reload. The panel in the plugin detail page is a live operations view: it shows whether the takeover is active and by which hook, how many characters the last assembly was trimmed from and to, and which providers are online (probed concurrently, with a real one-shot completion reporting latency and reply text per provider). If a local backend is not running, the panel offers a one-click launch: it probes the port first (never starting a duplicate), then starts the matching server (ollama serve, or a custom command you supply) detached and polls until the endpoint answers, clamping the wait to 3-600 seconds, reaping the child it just spawned if it never answers, and naming the paths it searched when it cannot find the binary. A stop button kills backends this plugin started (SIGTERM, then SIGKILL) and refuses to touch a server somebody else launched. When a model has just been downloaded and no backend exists yet, the panel can also scan the machine for .gguf files with hard caps (300 models, 20000 directory entries, Downloads only one level deep), start llama-server on the chosen one (skipping the .mmproj. vision projection and reusing a server that already answers), register the provider under the model id the backend reports, and run one real completion to prove it works. Before spawning anything it does a fit check that counts only the part of the file the GPU cannot hold: hostBytes = fileSize - VRAM, compared against max(total RAM - 2GB, free RAM). A file far larger than RAM is refused outright; a heavy host-side remainder is refused unless you confirm; the same remainder under mmap is only a warning, because the weights land in reclaimable page cache. A load-mode selector (auto / mmap / none / mlock / mmap+mlock) maps to llama.cpp --load-mode, and the default is whatever the engine picks. From 0.4.2 everything above is derived from the platform instead of assumed: home and data directories come from os.homedir() and the OS conventions (LOCALAPPDATA on Windows, ~/Library/Application Support on macOS, XDG_DATA_HOME elsewhere); backend binaries are looked up in the official install locations of each OS and can be pinned with OLLAMA_BIN, LLAMA_SERVER_BIN, VLLM_BIN or LMSTUDIO_BIN; model discovery walks DSH_LOCAL_MODELS_DIR, ~/models, ~/llama.cpp/models, ~/.cache/llama.cpp and the LM Studio data directory, with the WorkBuddy layout demoted to one candidate among them. VRAM is asked from nvidia-smi, then rocm-smi, then sysctl hw.memsize on Apple Silicon (flagged unified, so the same bytes are not charged twice), and any flip to a unified-memory budget is handled explicitly. Before building a command line the plugin runs llama-server --help once and keeps only the flags that build actually has, so older engines get --n-gpu-layers 99 and no KV quantization while newer ones get auto plus q8_0; anything it cannot probe is passed through rather than dropped. LM Studio is started with lms server start --port N rather than being lumped in with the engines that need a model argument, and vLLM gets its own --model example instead of llama.cpp syntax. Windows, macOS and Linux are all exercised in CI, including a windows-latest and a macos-latest job, because a cross-platform claim that has only ever run on one OS is not a claim. Two of those suites exist specifically to catch this machine's blind spots: one replays the whole suite with process.platform rewritten to each OS, and one deletes every advantage this development box has — no nvidia-smi, no rocm-smi, no sysctl, an empty PATH, a home directory with no models — and requires graceful degradation rather than a crash. That last one caught a real class of bug: several assertions hardcoded 'lms.exe', which cannot pass where the executable has no suffix. Measured on the reference machine: a 16.9GB 27.2B MoE with 14.5GB of VRAM holds a 285-691MB working set and sustains 76 tok/s after a 1000-2000 token warm-up, so file size alone is not a fit signal. From 0.4.3 the scanned models are shown as a table rather than a dropdown (one row per gguf, with its size and any fit warning), so you can see what is on disk before choosing. From 0.4.4 the delete button on the provider table removes the provider entry, not the weights file on disk. That distinction matters: what people actually want to clear out is the dead entry - one pointing at a port that never answered, one registered twice for the same port, one whose model file was deleted long ago - not a 15.8GB gguf. Deleting a config is undone by importing again; deleting a gguf is not. The removal goes through settings.mutate with an unset op on \['providers', id], because update merges layers: writing null there leaves a null provider key that the loader reads as a broken config rather than an absent entry, and only mutate/unset is implemented as Reflect.deleteProperty. The result is read back afterwards, because unset on a key that comes from a lower layer (the profile cordis.patch.yml or a home patch) does not delete it - it restores the inherited value, so the call succeeds while the entry stays. When the entry is still there the panel says so and points at the file to edit instead of reporting success, and a DSH whose settings service predates mutate() is told plainly rather than silently ignored. Deleting takes two clicks and clearing the confirmation is as easy as clicking another row; on success the online and baseURL caches are cleared and the tool surface is recomputed, since the provider list changed. The gguf itself is never touched by this plugin - the scanned-model table is display and selection only.VRAM reporting is driver-dependent. From 0.4.5 model discovery also reads Ollama's own storage: its blobs are named sha256-… with no extension, so extension matching could never list them. Weights are now accepted by extension OR by GGUF magic header (only dot-free files of at least 8MB, capped at 600 header reads per scan; obviously-not-weights extensions like .json/.md/.safetensors/.part are never opened). Ollama models are resolved from their manifests: the model layer is picked by mediaType application/vnd.ollama.image.model (never adapter/mmproj layers), the human-readable name is restored from the manifest path (…/qwen3/8b becomes qwen3:8b, …/latest stays bare), tags sharing one blob merge into a single row, and the metadata kv supplies name, quantization and context length. Adopting an Ollama model does not start a duplicate llama-server over the same weights: it confirms ollama serve answers on 11434 (launching it if not) and registers the provider under the tag the endpoint itself reports — the blob's sha256 name would never match anything, since 11434 lists qwen3:14b, not sha256-a8cc…. Two real-machine-only traps fixed on the way: existsSync('<file>/') returns true on Windows, so manifest-vs-directory branching now goes by parseable content instead of trailing-slash semantics; and the quantization label table had been copied from a stale ggml enum generation, labelling ggml_type 15 as Q5_K_S while ollama show reports Q4_K_M. Same-blob dedup is by (dev,ino), keeping whichever entry has a name. Verified on the reference machine: the scan goes from 1 row to 4 — qwen3:14b (8.64G) and qwen3:8b (4.87G) now appear with Q4_K_M badges — in 4ms; npm run verify exits 0 (launcher 57, portable 35, panel-host green including the linux/darwin simulated-platform runs); the new panel-host suite builds a real miniature Ollama store in tmpdir and points OLLAMA_MODELS at it, so results no longer depend on whether the test machine happens to have Ollama installed. From 0.4.6 the panel shows each model its real name instead of its file name, because some download tooling stores every weight under one generic file name, leaving the real name only inside the GGUF header. The scan now reads each file GGUF KV section - not a fixed-size head buffer, because tokenizer.ggml.tokens (hundreds of thousands of strings on a large MoE model) and the chat template push general.file_type past any window; and GGUF string arrays are interleaved (u64 length plus payload per element), so treating them as a length table walks the cursor off the rails and produces astronomically large reads. Reading and skipping are accounted separately: key names and types are always read, string values over 256KB and string arrays are only walked past without copying, non-string arrays over 64 elements are skipped outright, and a cursor past the file size stops the read, because an out-of-range position means the cursor is wrong rather than that a limit should be raised. The displayed name goes Ollama tag, then general.basename plus size_label, then general.name, then the file name, with the full string kept for the hover text since it usually carries internal build numbers; mmproj files are now identified by general.type as well as by name. A short read (truncated file) no longer feeds half a value into the parser. On the reference machine, reading one file header parses 44 keys in about 16ms, and a whole scan takes about 31ms. From 0.4.7 the smoke test no longer fails the first time and passes the second. The old code gave both request attempts one shared AbortController and one shared 30-second timer, so the total budget covered both rounds: a cold load of a large model burned it in round one, the abort left the signal permanently aborted, and the second round (the Ollama-style reasoning_effort: none retry) could not even be sent - which is exactly why the first click returned This operation was aborted and the second succeeded. Now each attempt gets its own controller and timer, a timeout retries once automatically (retryOnTimeout, on by default) because a first-request timeout almost always means the weights are still being read from disk, and non-timeout errors (refused connection, HTTP error) are not retried at all. Timeouts report themselves: the message says the wait exceeded smokeTimeout, whether a retry already happened, and points at the new smokeTimeout setting (seconds, default 60, clamped to 5-600) - port responding is not the same as tokens coming out, so it is a separate knob from launchTimeout. Unrelated: an example naming one specific model was dropped from the panel copy, since not every machine has that model downloaded, and the explanation now describes the mechanism instead. From 0.4.8 adopting a plain .gguf no longer registers an unrelated Ollama model. The port field used to follow the last click rather than the file: selecting an Ollama entry moved it to 11434, selecting a plain .gguf did not move it back, so a 11434 left over from an earlier click stayed. The server accepted it as given, probed the port, found Ollama answering and borrowed it (the skipped-launch log line about the port already answering is this path), and then the model id - derived from the file name - could not be matched against Ollama's /v1/models, where the fallback was to grab the first entry in the list. Embedding models tend to sort first, so the adopt registered a model the user never selected. Now the port follows the file: a plain .gguf landing on another engine's port (ollama 11434, vLLM 8000, LM Studio 1234) is moved to 8090 and the panel says so. Only ports belonging to a known engine are overridden - a hand-typed 9999 aimed at your own llama-server is left alone. The model-id fallback no longer guesses either: it takes the only entry when there is exactly one (a single-model backend may legitimately report a different id than the file name) and otherwise reports an error listing the ids it did see, rather than registering a wrong provider. From 0.4.9 the panel can no longer vanish. It used to be possible for the whole section to disappear from the plugin detail page while the host stayed up, the HTTP bridge kept answering, and no crash log appeared. The cause was a scope error: a helper was declared inside one component and referenced from another, which is valid JavaScript and passes a syntax check, so it only threw at render time - when the user pressed the scan button and React ran the state updater. The host wraps every slot entry in an error boundary whose fallback is an empty anchor, so the throw was contained silently: the panel was not uninstalled or crashed, it was replaced by an empty div. The helpers now live in the module scope both components share. The client half also gained its own test - a minimal React stand-in loads the real panel module, really calls the components, really runs the state updater at render time, and presses the scan button, so this class of error can no longer reach a release.
dsh-rice-mode
frisk325/dsh-rice-mode
Three-position toggle (off / this turn / always-on) for an "unlimited mode" that injects a "do not conserve tokens" instruction into the system prompt; the "this turn" mode reverts automatically on the kernel agent/status idle event.
companion-memory
root-7-khx/companion-memory
Roleplay companion memory for DeepSeek Harness: per-conversation editable persona documents (1-100 weight per entry, AI may refine them in small story-driven steps), per-speaker favorability and bond in multi-person chats, active memories with AI-scored vectors, a nine-emotion table updated every turn, an AI self-evolution ledger, recent 48-hour transcript search, companion-style conversation compression, sleep-time compute that pre-guesses the next topics while a conversation is idle, per-conversation tone cards learned from its own past replies, a habit-hint small-model bridge that writes the user habits it extracts into this conversation memory, a date-archived local knowledge base with auto-packing by turns or tokens, and a unified integrations page that lists curated plugins and downloads their GitHub releases. Its HTTP routes have no DSH access check, so a web page open in your browser could call them.
thu-tok-auto
overdustd7/thu-tok-auto
Fetches Tsinghua MadModel API tokens and renews them from the saved SSO session on a 20-minute timer in a windowless browser, writes the token into the DSH credential MADMODEL_API_KEY and creates or reuses the madmodel model provider; one manual Get captures that session through the campus login window.
dsh-kujira-persona
tests4gptplay/dsh-kujira-persona
Model-linked characters and a shared world for DSH, with separate editable personas and memories, optional AI diaries and sleep consolidation, Chinese/English controls, task-aware desktop pets, themes and usage statistics; updates preserve user customizations. Supports Windows, Linux and macOS; see the guide for each system's installation requirements. Diaries and consolidation consume tokens. Before installing or updating, we recommend closing visible desktop pets and bubbles; open files may cause installation to fail. Codex and OpenAI API account queries have been user-validated; other sources await user validation and feedback.
HwFee-dsh (usage-stats)
hwfee/hwfee-dsh
Usage panel on the blank-session hero: sessions, messages, total tokens, active days, peak hour, favourite model and a contribution heatmap, over all time, 30 days or 7 days.
dsh-skill-scope
dotdance/dsh-skill-scope
Lighten the load on your agent: one-click toggles for skills and MCP, so the context holds only what this turn actually needs. Skills and MCP servers you switch off are no longer injected into the model context — cleaner, and lighter on tokens.
EVA-Inspired-Theme
whatcannotbesaid/eva-inspired-theme
A Neon Genesis Evangelion-inspired skin for DeepSeek Harness: Title Card (light) and Central Dogma (dark) modes over one removable override layer, 55 measured color tokens, a wallpaper paint layer whose two surface alphas are solved against a WCAG contrast budget, a boot mask and a brand mark. It also rewrites the sidebar identity line, replaces the new-session heading, hides six official entries, and reshapes the composer control row. Unofficial fan work, non-commercial: the background pictures are fetched from third-party URLs at paint time, not shipped in the repository.
dsh-theme-herta
herta-st/dsh-theme-herta
A Herta theme for DeepSeek Harness: an ASCII boot opening ported from the upstream Herta project, a violet palette of design tokens, swappable wallpapers (four bundled, plus image URL, local import, solid colour or off) and a "Herta Appearance" page in DSH settings. Doubles as the theme that dsh-herta ships with.
DSH-appearance
lx-hmkk/dsh-appearance
A lightweight font and theme plugin for the DSH client: Latin/Chinese/code font pickers that list only installed families, and five official palettes (One Dark Pro, Dracula, Nord, GitHub, Catppuccin) applied through the official theme tokens, including code-block syntax colors. No bundled fonts, no dependencies, no build step. Requires DSH 0.2.0-rc.2 or newer.
dsh-usage-stats
jockjrop/dsh-usage-stats
In DSH Desktop, see the tokens, cost and speed of every request, compare spending by model and provider, merge different names for the same model, check balances and compare real model performance.
dsh-dashboard
weibaohui/dsh-dashboard
Usage dashboard: offline-scans session logs to chart daily/weekly/monthly tokens, estimated costs, model/tool/skill/command leaderboards, output speed, working hours and quality metrics; gridstack+ECharts cards are drag-and-drop composable, with custom formulas and AI arrangement (prompt round-trip import).
experts-management
weibaohui/experts-management
Expert manager: manage ntd-format experts and expert teams (plugin.json + Agent MD + skill sets), browse and install from a built-in market of 50+ experts, and run tasks in an expert persona via /expert-<name> without consuming model directory tokens.
dsh-secret-scrub
jkt-check/dsh-secret-scrub
Irreversible secret-scrubbing guard: rewrites access keys, bearer tokens, and private key blocks into `[REDACTED:<category>]` placeholders before they reach the session log and the model.
dsh-fuse
klrsl/dsh-fuse
Page-level UI generation for DeepSeek Harness: design tokens (theme.json) and code standards (code-style.json) drive a dsh-fuse fence that renders whole pages, with spec pre-validation, a theme contrast guard, a pixel-level style inspector (getComputedStyle), undo history, and a settings UI.
dsh-continue-on-limit
qwert702/dsh-continue-on-limit
Auto-continue for the dsh web GUI: when a local model hits its output-token cap and the max-tokens notice appears, the plugin automatically queues a short continue message so the reply keeps flowing — dual-source truncation detection, loop protection, and no UI.
dsh-ui-spec
yumimanji/dsh-ui-spec
DeepSeek Harness plugin that turns UI screenshots into implementation-grade web specs using OCR, deterministic geometry, scene graphs, assets, and render comparison.
dsh-usage-monitor
noirbright/dsh-usage-monitor
Settings usage dashboard: tokens, requests, output tokens, and cache-hit rate from session logs, with a stacked chart and a breakdown table by provider, model, or workspace.
dsh-model-picker
genius-alray/dsh-model-picker
Requires DSH 0.1.5-rc.2+. Dialog-style model picker for the DSH Web UI: fuzzy search, collapsible provider groups, favorites with a favorites-only filter, and a thinking-level dropdown, replacing the built-in model selector with official UI primitives and DSW tokens.
dsh-soma (plugin)
tlr18348452080-hub/dsh-soma
A body state for the agent that the agent cannot set directly: touch, temperature, fatigue, hunger, thirst, circadian rhythm and hormones are integrated as differential equations and reach the model only as one \[body] line. A single soma_do tool takes newtons, degrees Celsius and cm/s; there is no set(). Tiredness also lowers max_tokens, so a worn-out model writes less instead of saying it is tired.
dsh-bots
kadaliao/dsh-bots
Grok-style bots for DeepSeek Harness: create bots with their own avatar, persona, model, skills, apps, scheduled tasks and memory; they message each other, a dispatcher bot hands out the work, and a usage view counts tokens by bot, task and day.
dsh-plugins
kadaliao/dsh-plugins
Grok-style bots for DeepSeek Harness: create bots with their own avatar, persona, model, skills, apps, scheduled tasks and memory; they message each other, a dispatcher bot hands out the work, and a usage view counts tokens by bot, task and day.
dsh-injection-manager
iwhbiwhb/dsh-injection-manager
Shows and switches what a new DSH session injects before its first request (prompt sections, runtime contexts, skill catalog, tool schemas, workspace rules) with a content browser, cross-session first-turn trend, health checks and five entry points: panel page, conversation tab, right sidebar, /inject and a Settings row.