Plugins

Floating todo panel in the DSH Web UI — a draggable window that snaps to any screen edge, today/more grouping in the home tab, a separate all-items tab with uncompleted/completed filters and incremental loading, items that open read-only with an edit toggle in the corner, image attachments that zoom in place and other files that download, a pomodoro timer (15/30/45/60 or custom, with configurable rounds and break interval) that sends a Windows notification and waits for confirmation before the next phase, and five agent tools plus a /todo slash command for reading and editing the list.

FOFA cyberspace mapping inside the conversation: check the account quota, run FOFA queries and export the full result set into CSV/JSON/TXT files in the session workspace by paging through a time cursor. A CSV or TXT export stopped by its time budget or by an upstream error continues into the same file from the saved cursor, so rows already fetched are kept. Results render as conversation cards (query, match count, file path), and the API key is saved from the plugin page into the Harness credential store.

WEB ONLY: designed specifically for the Web profile, NOT Electron Desktop. Windows support means Web on Windows, not the Desktop application; installing into Desktop can break startup or restart. Refresh immediately after installation or open /login; HTTP 401 on an old page means sign-in is required, not installation failure. Only an uninitialized data directory displays the temporary password; existing passwords survive reinstalling. Initialize in a trusted network and change the publicly displayed initial password immediately; passwords are never printed in terminal logs. Web lock screen, password/OTP/Passkey, persistent sessions and wallpapers; Windows Web power services can be enabled from the page. Linux/macOS need the bundled controller; npm updates require a dedicated managed install. Stopping interrupts active tasks. Auth kernel adapted from xbzbing/dsh-auth-gateway (MIT); do not co-install.

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.

Right-click menus for DSH sidebar rows: on DSH 0.2.2+ both rows use the host menus — workspace rows follow the session grouping (new session, copy path and reveal, pinned open-with apps, an open-with submenu, bulk delete) and session rows extend the menu that opens at the pointer with pinned open-with apps, an open-with submenu and permanent deletion of archived sessions; entries stay in the right-click menu by default and can be opted one by one into the shared "…" menu. On 0.2.1 the workspace menu is drawn by the plugin. Adapted from miao-ge/dsh-sidebar-context-menu and 0imzero/dsh-workspace-menu; do not install it alongside either - all of them attach right-click menus to the same sidebar rows.

Opens chat.deepseek.com in the right sidebar pane from one row in the left sidebar, leaves the current conversation untouched, expands to fullscreen on demand, and shares the sidebar browser cookie jar; a bundled script makes the login survive restarts. Desktop (Electron) only; keeping the login across restarts needs the bundled PowerShell patch, which is Windows-only and must be re-applied after every DSH upgrade.

Phone-portrait layout shell for the DSH Web UI: the sidebar becomes an off-canvas drawer opened by swiping from the left, a persistent 56px action bar sits under the composer, and the settings panel plus the model and permission pickers are reflowed for touch. Ships native, merged and floating layout presets, all scoped to html\[data-dshm-mode="mobile"], so desktop, tablet and landscape keep the stock layout.

Turns an ordinary conversation into a guided study session, keeping progress and lectures as dated files that survive a restart. A Study tab shows how far you have got and which ideas are still shaky without writing anything, and a Lectures tab lists each course's lectures as draft, dated, or archived.

macOS notifications with the standard Glass alert when a reply finishes or DSH needs your input.

Saves tokens in DeepSeek Harness sessions by reading any file by line range or symbol mention instead of whole, and by bounding oversized tool results while keeping a recovery path. Pruning happens at the source, not after content reaches the context. Measured 43.9% fewer prompt tokens in a modelled 30-step session.

Shows a dimmed suggestion of the user's next message in the empty composer after a settled turn; Tab accepts it into the draft and Escape or typing dismisses it.

Adds a button to the composer that restates your draft, then asks a few questions with clickable options — every question offers an "I don't know, you decide" escape — and only then rewrites the draft into a request an agent can start on. The whole flow can be finished without typing, and it never sends for you.

Power Mode typing effects for the DSH web composer: caret particles and glowing character echoes, a heat ramp from amber to red with rising embers as the combo grows, a combo counter over visible glyphs, and a synthesized key tick with a heavier Enter impact.

Windows only: a night-and-moon themed floating button (a sleepy DeepSeek-chan whale mascot, awake/sleeping artwork) in the bottom-right corner arms the plugin — click it to power the machine off once every root Agent session has gone idle, and click it again to disarm. The button is draggable (position remembered in localStorage), automatically dodges other plugins' corner popups, and survives the frontend re-rendering it away. When the batch finishes a 60 s cancellable countdown starts (configurable 10–600 s) with the remaining seconds on the button and a matching banner: clicking Cancel runs shutdown /a and disarms the mode, and any new session starting during the countdown aborts it. A right-click menu on the button switches the shutdown condition between only-on-success and shutdown-regardless, persisted across restarts in $DSH_HOME/deepseek-chan-sleep-shutdown.json. Zero runtime dependencies.

Edit a sent message in place and resend it as a forked chat, keeping the original: change the text, remove, reorder or add images and files, with session image limits checked before sending (the first message opens a new chat in the same workspace). Also @-references pending images by file name, drag-reorders composer attachments, copies messages together with their images, and adds copy, download and drag-to-composer for images in the conversation.

Native generation-time Mixture-of-Agents: every turn, reference models run the same session context in parallel as pre-run drafts and one aggregator executes with full tooling; config-as-model recipes, F mirror / R trimmed views, enforced subagent gating (dual-shape session indexing), derived per-recipe vision (vision refs relay image reads to text-only aggregators), tiered cadence with separate steering and patrol rosters plus frozen reuse between beats, degeneration guard dropping degenerate reference output, anonymized model face (failure notices carry no provider identity), and cache-friendly guidance placement (the rehearsal record is pinned where it was produced instead of appended to every request, so the cached prefix covers it and each step pays only for what was added since - on by default across every cadence).

Repairs dsh-chaos-fixer-crasher by disarming its exit trigger before it fires, keeping the four-plugin chaos chain stable. Warning: it only disarms this specific crash, not arbitrary uncaught exits. Joke plugin.

United Nations theme for the DSH Web GUI: UN blue over paper white, the official emblem, a slow rotation of credited UN peacekeeping photographs behind the conversation, Charter and UN-facts ribbons, and a 7,030-language picker (127 languages with a complete skin translation) that leaves every existing control untouched. Browser-side only; installers for Linux, macOS and Windows.

A draggable task-tree panel inside the chat page: capture tasks by voice or text, drag to reorder, edge-snapping collapse, and a plain-text queue file you can edit or version yourself.

Folds the built-in subagent dropdown into History / Today / Previous / Current groups by usage batch, keeping the current batch expanded; the original header is left untouched.

Codex-style drag-to-reference for DeepSeek Harness: drag a conversation row from the sidebar onto the chat area and it becomes a native session-reference chip in the composer draft, so you can keep typing around it before sending. Drag a workspace (folder) row instead and the draft gets that folder's path, so a whole directory can be handed to the model. The target session is matched by session id rather than by title, so duplicate or blank titles cannot insert the wrong conversation. The reference cap comes from the host (3 by default) and is read at runtime instead of being hard-coded, so a host-side raise is followed without changing the plugin. If the chip cannot be inserted, the plugin falls back to sending a reference-only message instead of failing silently.

Adds a ✦ button below the DSH composer — and on the new-conversation screen — that rewrites the current draft into a more complete prompt with a single call to the configured default model, keeping the original intent and language, and replaces the draft in place without sending it.

Translates PDFs with BabelDOC through an OpenAI-compatible endpoint while keeping the original layout, and returns both a bilingual and a translated-only PDF. Requires a one-time runtime setup and an OpenAI-compatible API key.

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.