Skip to main content

Plugins

Browse, filter, and install DeepSeek-Harness plugins.

79 plugins found

L

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.

22 days agoModels & ProvidersMIT
G

dsh-auto-rerun

good-saltfish/dsh-auto-rerun

Automatically retries model failures (drops, timeouts, rate limits, 5xx) in-turn. Retry count, delays, backoff, fallback and more are configurable. Rate-limit backoff is capped and jittered, server retry hints are honored, and optional retry-until-success keeps going within a time ceiling. Every option is editable from the plugin config or the sidebar panel: retry count, initial and max delay, backoff factor and jitter, rate-limit delay window, retry-until-success ceiling, fallback trigger and switch cap, dead-route blacklist TTL, log file. When a route keeps failing it can switch to a pinned, allowlist-filtered, or catalog-picked alternate model route, with a 30-minute dead-route blacklist (auth failure, subscription expired, billing) and a stale-assembly guard that keeps retried requests on the switched route. Ships a right-sidebar panel with live retry state, route labels and incident counters, Chinese and English locale, and memory-only logging by default.

216 days agoModels & ProvidersMIT
J

dsh-question-timeout-protection

julardepick/dsh-question-timeout-protection

A plugin for dsh: once ask_user_question times out, the content you entered will remain in the UI or model context.

2yesterdayUI EnhancementsApache-2.0
Q

dsh-ask-guard

q1hangl/dsh-ask-guard

Cooperative timeout guard for ask_user_question: lost or unanswered questions resolve as a structured ASK_TIMEOUT instead of hanging the turn forever.

23 months agoTools & CapabilitiesMIT
B

dsh-free-subagents

boldandigital/dsh-free-subagents

Route delegated subagents and Agent Teams over free-tier models: choose a pool of approved free routes once, and every child request fails over to the next free model with a persistent ledger of quota, rate-limit and timeout parks.

18 hours agoModels & ProvidersMIT
L

dsh-sandbox-approval-modal

londonmoon/dsh-sandbox-approval-modal

A top-most sandbox-escalation approval modal for DeepSeek Harness with a FIFO queue, one-shot allowed-once grants, reject-with-comment, and host-enforced timeout.

13 days agoSecurity & PermissionsBSD-3-Clause
C

dsh-use-codex

chaorenzeng/dsh-use-codex

Delegates independent tasks from DeepSeek Harness to the local OpenAI Codex CLI and returns the result to the current conversation. Provides four tools (task execution, status check, danger-mode toggle, enable/disable switch), enforces a real-path working-directory whitelist with symlink-escape protection, defaults to Codex workspace-write sandbox mode, passes the task body as a single argv argument to avoid shell injection, kills the whole Codex process group on timeout or cancel, and caps output length. Requires Node.js >= 20 and a Codex CLI build supporting codex exec.

110 days agoTools & CapabilitiesMIT
L

DSH-Mem0

light-cheng/dsh-mem0

Mem0-backed long-term memory for DeepSeek Harness with managed Platform, self-hosted server and local OSS backends; visual settings, automatic recall (from the first turn) and capture, editable memory summary, token budgets and weekly memory hygiene that only deletes memories this plugin wrote. Captured text is redacted; the memory summary is a stable system-prompt section (plugin:dsh-mem0:profile) and recall a separate mem0:recall block appended last, so the cached prompt prefix stays stable; recall and capture have timeouts (recallTimeoutMs, addTimeoutMs). With dsh-strata installed, recall size follows its strataBudget allowance, and the optional captureStrataFolds setting (off by default) also saves its fold summaries as memories.

13 days agoMemoryMIT
W

dsh-jj-panel

wzciczw/dsh-jj-panel

Jujutsu (jj) version-control panel in the DSH web GUI: a Jujutsu tab renders the jj log change graph with change details, changed files and file diffs, a revset input, and jj-native history edits (drag-to-rebase, split by file, describe, squash, abandon, undo, push); it reads jj directly rather than through git, and every command runs as an argv array through a whitelisted host service with confirmations, timeouts and output caps.

124 days agoGit & Code ReviewMIT
W

dsh-model-connection-test

willmylife/dsh-model-connection-test

Adds a connection-test panel under Settings -> Models: reads the deployment's model catalog and probes each configured model with one minimal chat round-trip (a short "ping" with a small output cap) through the Host LLM service, verifying connectivity with a real request. Each row reports success/failure with the harness failure code (NO_ADAPTER, MISSING_CREDENTIAL, AUTH, RATE_LIMIT, TIMEOUT, ...) plus first-token/total latency; single-model and concurrent test-all (3) are supported. Credentials are resolved inside the Host process and never reach the browser; probes are independent per model behind a same-origin-fenced route with a body-size cap. Pure JS, no build step, MIT.

117 days agoModels & ProvidersMIT
B

dsh-routing-forensics

bushizhanghan/dsh-routing-forensics

Forensic analysis of past model-routing decisions. Replays each session, classifies every turn's task complexity from its request shape (deterministic, no LLM), and reports two kinds of waste: an expensive model on cheap work (the difference is measurable), and a cheap model on expensive work that failed on its own merits — where every retry re-sends the whole context, so the downgrade cost more than the upgrade. Distinguishes provider-capacity retries (RATE_LIMIT/TIMEOUT/TRANSPORT) from model failures so it does not cry wolf. Derives routing rules from the evidence, priced against a built-in per-million table.

19 days agoUsage & BillingMIT
W

dsh-tandem

wuuuuu123-32/dsh-tandem

Runs a second agent: hands independent subtasks to a local Codex CLI in parallel (up to three at a time), and asks it to review a design or a plan before the user sees it, returning conclusion, evidence, counterexamples and a verification step. A dispatch missing its timeout or its acceptance check is rejected by the plugin, not by the model remembering to check.

13 days agoWorkflow & AutomationMIT
R

dsh-gitbash-new

riverzhou/dsh-gitbash-new

Windows bash tool for DeepSeek Harness: replaces the preset pwsh with Git for Windows or MSYS2 (UCRT64/MINGW64/CLANG64/...) bash while keeping the official terminal contract — workdir, background jobs and timeout promotion, result markers, terminal cards — and adds a settings row that discovers and validates Git/MSYS2 installations.

12 days agoTools & Capabilities
I

dsh-mcp-manager

it-glen/dsh-mcp-manager

MCP server manager for DSH: add, edit, enable and disable Model Context Protocol servers from a panel, with stdio and streamable-HTTP transports, per-server tool-call timeouts, live connection and tool-count status, and persistence to a project- or home-level mcp.json. Maintained fork of @wingsky-1/dsh-mcp-manager.

126 days agoTools & CapabilitiesMIT
Q

dsh-model-failover

qgg4588/dsh-model-failover

Model failover for DeepSeek Harness: pick one or more providers and the pool becomes every model inside them; when a request fails (rate limit, quota, timeout, server or transport error) the next untried model in the pool retries automatically, and the original error surfaces once the pool is exhausted. Configure from a seat beside the composer model picker or the model_failover tool, with no restart needed.

125 days agoModels & ProvidersMIT
X

dsh-model-endpoint

xxsttb/dsh-model-endpoint

Publishes whichever model the profile has selected in chat or on the Models page as an OpenAI-compatible HTTP endpoint: /v1/models, /v1/chat/completions (streaming SSE and non-streaming), /health. A request may name any routable model; one that omits `model` gets the current chat selection, so switching model in chat switches the endpoint. Switched on, it can also lend the model the Harness tools switched on individually, run those calls through the Harness tool runtime and feed results back up to a round limit. The listener is the plugin's own, so 127.0.0.1, 0.0.0.0 and a NAT-forwarded port all behave the same with no change to the request path. Settings cover bind address, port, API key with a generator, fallback provider and model, reasoning effort, CORS origin and timeout.

1last monthIntegrations & RemoteMIT
M

dsh-serper-search

minokun/dsh-serper-search

Complementary web-search provider: every web_search merges the DeepSeek Search API (primary) with serper.dev Google results (secondary) at a fixed 2:1 ratio. Both sides are direct HTTP retrievals with no auxiliary model turn: URL-deduped sources, long DeepSeek summaries clipped into snippets, per-backend timeouts and graceful degradation to whichever backend answers.

117 days agoTools & CapabilitiesMIT
K

dsh-qoder-link

k1-h2/dsh-qoder-link

DeepSeek Harness LLM adapter for Qoder CN subscriptions, built on the official @qodercn-ai/qodercn-agent-sdk. Reuses the local qoderclicn login, and exposes a fully configurable settings page (per-model context window, session pool, timeouts, model deny l

1last monthModels & ProvidersMIT
1

dsh-wsl-power

173787247/dsh-wsl-power

Windows power state from WSL: active power plan, all schemes, battery, sleep timeout and hibernation availability.

13 days agoDev & Plugin ToolsMIT
M

DSH-Rtk-rewrite

mengqi1436/dsh-rtk-rewrite

Replaces the web profile's pwsh shell executor with a subclass that rewrites every command through `rtk rewrite` before execution for token savings, running the original command unchanged on any rewrite failure (no equivalent, missing rtk, timeout) and keeping sandbox semantics intact.

1yesterdayDev & Plugin ToolsMIT
B

dsh-plugin-precheck

baqif2/dsh-plugin-precheck

Pre-install compatibility gate for DSH plugins: locks the resolved version, checks peer/engines declarations, dry-installs into an isolated DSH_HOME and boots it with a timeout, blocks on any failure, and restores the profile if the real install fails. Optionally intercepts dsh-market install/update clicks. Compatible with DeepSeek Harness 0.2.x.

1last monthSecurity & PermissionsMIT
A

dsh-model-channel-manager

arcaneorion/dsh-model-channel-manager

Model channel manager: several provider/model candidates share one virtual route (roundrobin/<group>) with failover, retry, timeout and cooldown; a settings tab edits providers and groups, runs per-model speed tests and real single-model request tests, and a seven-day health panel aggregates latency, failures and upstream token usage (input, output, cache and reasoning).

12 months agoModels & Providers
B

dsh-semgrep-sast (bundle)

baiiduu/dsh-semgrep-sast

Workspace-scoped Semgrep SAST tool for DeepSeek Harness with a managed Windows x64 runtime, structured bounded findings, cancellation and timeout controls, and user-approved sandbox escalation. Installs from npm as @aaub-software/dsh-semgrep-sast.

1last monthSecurity & PermissionsMIT
A

dsh-ark9canvas

aik358/dsh-ark9canvas

Image generation workbench and agent tool for DSH: one tool (ark9_generate_image) paints text-to-image and image-to-image via any OpenAI-compatible API. Channels are entirely user-configured - no bundled endpoint, you bring your own baseURL, key and model name (manual input supported). Agent-initiated generations are approval-gated by default: requests queue in the floating panel and nothing bills until you approve; approval is non-blocking (the agent replies "awaiting your approval" immediately, and results, denials or timeouts are injected back so the agent reports them in the next turn). Floating glass workbench with five tabs (Generate / Approvals / Prompts / History / About), aspect-ratio grid with quality-budget + 16px-alignment sizing, transparent background, batch up to 10 images as aggregated sub-tasks, prompt library with custom JSON sources fetched through a host proxy, persistent generation logs with one-click retry, multi-channel aggregation with per-channel model fetching, bilingual UI (zh/en), stroke-SVG icon set, config import/export. Registers as a Better Sidebar tab when dsh-better-sidebar is installed; stacks above the dsh-cua FAB when both are present.

12 months agoTools & Capabilities