Plugins
Browse, filter, and install DeepSeek-Harness plugins.
1010 plugins found
dsh-gatekeeper
huajikano/dsh-gatekeeper
Inspects the model's streamed reply before it is committed and either discards it or holds it for review when it reads as a refusal, so that text does not stay in the session context. Detection combines built-in phrase rules (an opening refusal, a capability denial, policy wording, first-person intent, refusals quoted in reasoning) with a plain-text word list and optional regular expressions. Interception can sit at the llm/stream waterfall, at the HTTP fetch layer before encryption (not TLS decryption, and it cannot tell which session a call belongs to), at both, or in a preview mode that shows a transient preview while the original chunks stay in memory until the whole reply has been checked. The same plugin also guards tool calls on tools/pre-execute: a command that deletes a drive, filesystem or user-profile root, formats or repartitions a volume, empties the Recycle Bin, fetches and executes a remote script, or writes to a raw device is denied, and a recursive delete outside build directories is recorded by default. Each tier can be set to deny, ask, record, or off, and the disk guard has its own switch. Both halves are heuristics: they can miss phrasings, the disk guard sees the arguments of a call rather than what a running script does, and they fail open.
dsh-until-done
kunkun24/dsh-until-done
Keep working when a turn ends without finishing, and only stop once the reply ends with a completion keyword.
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.
reminder-dsh
kes1ke/reminder-dsh
Native Windows completion and approval notifications for DeepSeek Harness, with session titles, optional AI summaries, quick reply, and an opt-in taskbar completion badge. Toasts carry the original session route, subject to Windows activation. The badge requires PowerShell 7.
dsh-replay-reset
popcorey/dsh-replay-reset
Manually resets old AgentRouter Responses replay state for chats blocked by upstream resource-binding errors while preserving history; supports DSH 0.2.2-rc.1.
aurabot-qq-mac
geng-zy/aurabot-qq-mac
macOS (Apple silicon) fork of aureth-labs/dsh-aurabot-qq: the same QQ persona bot (chat, voice notes, stickers, memory, song covers, modules) with the Windows-only code removed. It connects to a OneBot v11 gateway such as SnowLuma running in OrbStack or Docker, falls back to the macOS `say` voice, sends files over 30 MB to a containerised gateway through a loopback-only temporary download link, and installs the song-cover environment (Applio, audio-separator, uv, Rubber Band, separation models) at pinned versions. It keeps the upstream package name and plugin id (dsh-aurabot-qq), so it replaces upstream and cannot be installed alongside it; use upstream on Windows.
dsh-plugins (dsh-opencode2api)
zhiqiangme/dsh-plugins
Runs the standalone opencode2api backend (jasonxu114514/opencode2api, Windows x64/ARM64, no Docker or OpenCode CLI) with DSH and on exit stops only the process it launched; its settings card installs, updates or rolls back the backend from checksum-verified GitHub releases, copies the initial admin password and syncs /v1/models into an opencode2api provider with context, vision and reasoning efforts, re-syncing every 10 minutes. Frozen archive: 26.10.31 is the last version, and the author points to dsh-magpie as the replacement.
dsh-big-fish-speak
chaowei08/dsh-big-fish-speak
Makes the dsh-pet desktop pet read model replies aloud in its own configured voice. Newer dsh-pet gets the whole text in one request (up to its 500-char cap); older versions automatically fall back to 60-char chunks. Each piece is verified against the pet log to confirm it actually played.
dsh-history-fold
mili1360/dsh-history-fold
Replaces the oldest span of a session's visible history with one message carrying an extractive index and the path of a lossless local archive, so each request sends fewer tokens.
dsh-breeze-tts
eslaykey/dsh-breeze-tts
Breeze Blue text-to-speech for DeepSeek Harness: agent tools, reply playback, voice design and a settings page.
dsh-content-search
appthin/dsh-content-search
Searches the whole conversation — not just the loaded window — from a button in the conversation header, opens the results in the right sidebar, paints every keyword yellow in both your messages and the replies, and jumps the conversation to the hit. History search runs on the host log, so it still works when the full-text index is disabled.
dsh-plain-talk
lizilong1993/dsh-plain-talk
Composer toggle that shapes replies: conclusion first, inline quotes, no cross-turn references.
dsh-watchlist
hongcha-dsh/dsh-watchlist
Follow list above the sidebar conversation list: a status dot per conversation (unread, waiting for you, running, failed), newest reply first, running ones last, and a follow toggle beside the model selector. Marking a row unread lifts it to the top, opening one expands and scrolls to its row in the Workspace list below, a finished turn always counts as unread until the conversation is opened, and a conversation whose delegated subagents are still working keeps counting as running until the last child reports back. Its HTTP routes have no DSH access check, so a web page open in your browser could call them.
cyber-im-avatar (dsh-plugin)
buger-gif/cyber-im-avatar
Sidebar IM stand-in for DeepSeek Harness. Reuses your logged-in Feishu/Lark, WeCom and DingTalk CLIs (DingTalk is enabled from the panel) to poll only the chats you watch; an @-mention or a new message wakes the bound DSH session, which reads context, researches and drafts a reply in your voice. Each watch has its own draft-approval / auto-send mode and a three-state attention level, and new groups are triaged by the agent. Blocked chats are never fetched, message bodies stay on this machine, and the plugin opens no port of its own.
dsh-littlestar-anything-voice
theskywilldawn/dsh-littlestar-anything-voice
Click the chibi mascot and it reads the assistant reply aloud through your local IndexTTS 2.5 install. One character = one reference clip; a missing portrait just shows a placeholder. Ships no voices or artwork — the library lives on the user's own machine. Requires a local IndexTTS 2.5 install.
dsh-tool-call-leak-fix
irwdone/dsh-tool-call-leak-fix
Blocks a tool call whose arguments lost their boundary, keeps a closing turn open when a step silently never ran so the model can re-issue it, and replaces damaged output on the model-visible surface after a fork. Zero dependencies.
dsh-soul-evolve
xiachuyoufeng/dsh-soul-evolve
Persona state and slow evolution for DSH: five state dimensions (valence, arousal, energy, intimacy, focus) move after each turn and decay toward baseline over time; four scene types (work, tech, support, chat) modulate reply length and warmth without changing the state itself; the Soul panel keeps a persona card plus weighted evolution entries and injects them into the system prompt persona-first under a 4000-character budget. Ships a browser panel with a persona card editor and a live state view (the five metrics plus the last six events), appends a JSONL trace to disk, and seeds a usable default persona on first install.
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.
CatDogTrace (dsh-plugin)
xiebaiyuan/catdogtrace
DogCatTrace attribution parody as a right-sidebar tab: draw three integer challenges, paste each model reply into one box at a time, read the 24-breed dog and cat ranking.
dsh-dingtalk-self
mineagle/dsh-dingtalk-self
Replies to DingTalk direct messages as your own account — reads the bound DSH workspace, writes the answer, and sends it back through the dws CLI.
voice_for_dsh (voice-for-dsh)
junarch/voice_for_dsh
Reads each reply aloud in natural spoken language, summarizing structured content such as tables and code blocks, with an adjustable script length — from detailed narration close to the source length down to a minimal briefing. Engines: system TTS, the DeepSeek TTS API, Doubao (dialects and expressive delivery), or the local Kokoro model, which synthesizes offline in-process. It also provides three session sounds — a repeating tone while the agent works, a completion tone at the end of a turn, and a question tone when it needs your answer; built-in tones can be switched and custom audio uploaded, with files kept on the local machine.
dsh-dsml-guard
jessicaisleep/dsh-dsml-guard
Strips DSML markers that leak into assistant prose, and resumes a turn when a "continue" genuinely helps — a marker-only leak that produced no output, a single-reply output token cap (max-tokens), or an error the host's own retry policy never covers; codes the host already retries inside the step (EMPTY_RESPONSE, SERVER, TIMEOUT, TRANSPORT) are handed to it by default, and setting skipHostRetriedCodes: false restores the plugin's own resume. Codes upstream miscategorises are split by message text, and consecutive resumes stop at 3 until a normal reply resets them; quota, rate-limit, auth and context-window errors, and user aborts, never resume. Requires DSH 0.2.1-rc.2 or newer, below 0.3.0, and Node 20 or newer.
dsh-onebot-channel
danshuyue/dsh-onebot-channel
Connects QQ private chats and groups to DSH over OneBot 11, adding qq_send and other tools so the agent can reply, send stickers and wait for follow-ups.
dsh-artifact-linker
loyalchiiina/dsh-artifact-linker
Adds a resident system-prompt rule asking the assistant to report every file it produces as a full absolute path (a soft rule, not enforced), and replaces the official context menu on a file link with eight actions: open in the sidebar, open in a floating window, open with the default app, reveal in Explorer, save a copy elsewhere, copy the file, copy its path, and add it to the chat box, so a result reaches a friend in one click (copy-file and save-as are Windows only). Adapts DSH kernel 0.2.2-rc.1.