Plugins
Browse, filter, and install DeepSeek-Harness plugins.
2731 plugins found
dsh-locator
perrylink/dsh-locator
Whole-disk file-name search for DSH with a self-built cross-platform index, agent tools and a Web panel; no Everything required. It indexes your home directory by default; the first content search also builds a local full-text index (SQLite in the platform cache directory) of text and source files there (by extension, up to 1 MB each, 200,000 files) and of your recent DSH session messages; locator_export writes a SQLite copy of the index to any absolute path it is given, replacing an existing file when overwrite is true. The engine is a prebuilt Go binary shipped for six platforms (source in engine/), and the plugin makes no network requests.
session-stickies-manage-your-attention
juliiiiiiiu/session-stickies-manage-your-attention
Session-specific sticky cards with optional, explicitly authorized agent follow-ups. Targets the DSH desktop/web profile (protocol 0.2.2-rc.1); installed from GitHub.
dsh-hoi4-modding
kiant98/dsh-hoi4-modding
Hearts of Iron IV modding bundle for DeepSeek Harness: registers 15 skills covering focus trees, events, decisions, ideas, characters, technologies, localisation, GUI and error.log triage, and mounts 20 local MCP tools (a Python stdio server) to scaffold content, statically validate mods, build a SQLite FTS5 knowledge base from the vanilla game, and preview focus trees offline. Ships a 163-document reference library; requires DSH >=0.2.0 <0.3.0 and Python 3.9+ on PATH (standard library only, no game assets).
dsh-agent-teams
chenxu233/dsh-agent-teams
AgentTeams for DeepSeek Harness — an independent design branch of NanmiCoder/dsh-agent-teams; do not install both, since both register the same `agent-teams` row. Multi-agent team collaboration with a captain and members, tasks that carry dependencies, and messaging, all driven by natural language, plus a tree monitor in the web GUI.
dsh-caveman
stevenwang15/dsh-caveman
Apache-2.0 port of the caveman skill family (JuliusBrussee/caveman) into DeepSeek Harness as 14 native skills — three terseness levels (caveman, ultracave, megacave) plus commit, review, help, stats, an in-session compress whose gate rejects a dropped code block, heading, URL, negation or env var, or a result that did not shrink, and six workflow-discipline skills; commands, paths and error text stay verbatim.
dsh-sidebar-suite
yonan1269/dsh-sidebar-suite
Adds two switchable sidebar layouts (tabs and overlay) plus a no-override official default, an always-expanded session list, disposable temporary sessions and a third-party tab API to DeepSeek Harness, with default-layout and feedback controls in the plugin detail page.
agentkit-studio
kingsoftcloud/agentkit-studio
AgentKit Studio product plugin with an independent TypeScript provider for the official DSH host.
dsh-theme-cyrene
linxi-1105/dsh-theme-cyrene
A free, unofficial Cyrene-inspired theme for DeepSeek Harness with three palettes, a splash screen, and a chat watermark. Code is MIT; character fan art is excluded from MIT.
dsh-baidu-netdisk
justinzyc271828-sys/dsh-baidu-netdisk
Connects DeepSeek Harness to Baidu Netdisk's official MCP service for file search, file metadata, and storage quota. Requires Baidu account authorization after installation.
dwguard-dsh-hmos
dogwind221/dwguard-dsh-hmos
HarmonyOS delivery orchestration for DSH: freezes a four-dimension acceptance pack whose measurable criteria carry machine-checkable assertions, runs an adversarial pre-review before dispatch, hands the work to DevEco Code in Goal mode, then audits the delivery with independently gathered evidence (screenshots, UI tree, logs, hashed evidence bundle), recomputing assertion verdicts against measured values and filing rework orders; includes a cited rule catalog, a user-confirmation state and a gate calibration ledger.
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-shark-bg
iemgnod/dsh-shark-bg
ASCII sharks and schooling fish swim behind the DSH Web interface, with a settings page for veil opacity, shark count, size, fish density, swim speed and grid strength.
dsh-gpt-image-playground
southcul/dsh-gpt-image-playground
Image generation and editing for GPT Image, Gemini (Nano Banana) and Grok, with official, OpenAI-compatible, api.toolify-ai.com and custom HTTP providers. Text-to-image, reference-image editing, canvas annotation, mask editing and Agent multi-turn chat, with images and history kept on your machine. The workbench UI is the MIT-licensed GPT Image Playground v0.9.0 build by CookSleep, bundled unchanged; that upstream is a standalone web app, not a DSH plugin.
dsh-autocad-bridge (bundle-autocad-ui)
cmm00100/dsh-autocad-bridge
Drives the installed AutoCAD over COM and exposes it to the agent as 18 mcp__autocad__ tools: connect and status, cancel the current command, list registered ProgIDs, list or create documents, save, create and list layers, draw lines, circles, rectangles, polylines, arcs and text, list entities, zoom to extents, and pass any native command through. A sidebar panel owns the Python bridge process — start, stop, restart, dependency bootstrap, live logs and crash recovery — and resolves every path at runtime, so the same bundle serves both the web app and the desktop app. Windows only, Python 3.10-3.12, AutoCAD 2022-2024 documented (2024 is the default; other versions require setting CAD_PROG_ID to the matching ProgID).
dsh-usage-wallet
yeoksin/dsh-usage-wallet
Unofficial fork of dougen/dsh-deepseek-usage (baseline v0.16.0): sidebar balance and quota card for the active model provider, covering DeepSeek, Moonshot, Z.ai, OpenRouter, Kimi, MiniMax and OpenCode, plus Alibaba Cloud voucher balance for qwen-token-plan (BSS OpenAPI with AccessKey) and Xiaomi MiMo console balance with token-plan usage (console cookie). Ships a provider query-status page and per-token unit prices. Do not install it alongside upstream dsh-deepseek-usage: both render the same sidebar card.
xander-zaozao-xiaban
gx18391303625-afk/xander-zaozao-xiaban
Xander Amazon seller-operations toolkit for DeepSeek Harness: 33 native xander_* tools and 7 skills for competitor research, keywords, listings, ads, replenishment and profit, with exact proposals, explicit approval and read-back verification for Amazon changes.
dsh-cngtn-pond
cngtn/dsh-cngtn-pond
Paper-and-ink theme with warm surfaces, hairline strokes, near-square corners and a living ASCII koi pond behind the shell; the theme sets no font family, so the shell's own stacks (or your Settings choice) apply.
dsh-lead-mode
xiebaiyuan/dsh-lead-mode
Lead-mode agent preset plus two runtime skills for the official Agent Teams runtime: a delegate-wait-verify working method, and a wrap-up digest that merges several members' output into one file. Ships no team runtime of its own.
dsh-lark-agent
nirvana-jie/dsh-lark-agent
Reach your DeepSeek Harness assistant from Feishu / Lark: a dsh plugin built on the official Lark CLI
dsh-workplan
abbaababa/dsh-workplan
Durable task-decomposition plans for DeepSeek Harness with a self-learning loop: every department subtask runs in three ordered phases (plan, execute, verify), the verify phase must submit a retrospective of problems, detours and resolutions, and that retrospective is written to a permanent lessons file, ranked against the new subtask and injected into later dispatches so the same mistake is not repeated. Adds a read-only dsh_workplan_lessons tool (11 model tools in total), a progress panel drawn as a ring, a stacked bar and per-stage steps, and a public ISO/IEC 15408:2026 security self-assessment with zero runtime dependencies.
dsh-incident-journal
qw2007/dsh-incident-journal
Five skills for a reviewable engineering incident journal: record an event only when a counterfactual can be written, quiz yourself from the existing entries, audit the journal for decay, ingest outside material together with its provenance, and rename a layout directory while rewriting every reference to it.
dsh-lane-bridge (dsh-lane-bridge)
cnranger1124/dsh-lane-bridge
Multi-channel IM bridge for DeepSeek Harness: each QQ (SnowLuma / OneBot v11, official QQ Bot) or Discord group and private chat becomes its own agent session, bound to a role card whose domain carries knowledge bases with draft approval, shared memory, a read-only GitHub reader and group trigger rules; two built-in personas (customer support, group chat), unified push with broadcast, and human hand-off.
dsh-workspace-memory
libraqg/dsh-workspace-memory
Explicit, project-scoped memory handoff: when you open a new conversation, one model call that you confirm curates the project's earlier conversations into a bounded brief and injects it into the conversation's first step. No silent background calls; a Memory panel shows the brief and lets you edit notes.
dsh-mnemosyne
i11ustrious/dsh-mnemosyne
Layered long-term memory for DeepSeek Harness: each turn injects a persona frame plus a relevance-ranked memory index while full transcripts stay on disk and are expanded on demand, so the vault never floods the context window. Facts are slot-keyed, so a changed mind retires the old value instead of recalling two contradictory answers. The persona frame is derived from memory by the model itself, behind three anti-pollution gates (rejecting development chatter, blocking echoes of its own injection, and merging restatements by shared content words). Periodic consolidation merges near-duplicates, rolls long topics into digests, and archives low-value old entries by tagging only — raw text is never deleted. The append-only store is compacted in place (measured 4.7 MB to 355 KB). Zero third-party dependencies, no build step. Note: per-turn injection inevitably lowers prompt-cache hit rates, so this plugin is not recommended for work scenarios.