Plugins
Browse, filter, and install DeepSeek-Harness plugins.
149 plugins found
dsh-image-gen
leemancheung/dsh-image-gen
GPT Image 2 `image_gen` with Codex subscription OAuth by default or explicit API-key mode: developing card, up to three live API partials, durable attachment replay/lightbox/download, text-only model output, and bounded credential-safe requests.
dsh-opencli
ikeasven69/dsh-opencli
Logged-in browser automation with 170+ site adapters as deterministic one-shot commands (requires the OpenCLI CLI, npm i -g @jackwener/opencli, and its BrowserBridge browser extension) — action-evidence verification (no silent failures), site knowledge & health cards, watch monitoring, trace replay, provenance reports, and a write-approval gate.
dsh-record-replay
humblebanana/dsh-record-replay
DeepSeek Harness record macOS desktop workflows by demonstration and turn them into agent skills (open-record-replay skill + orr_* tools)
dsh-plugin-rollout-scout
spookysandwich/dsh-plugin-rollout-scout
Detects which conversation model an account is being served: launches concurrent throwaway probe conversations, scores each streaming chain-of-thought by how its paragraphs open, and cancels probes that read as the older model within seconds. Includes an offline self-check that replays 13 hand-labelled chains-of-thought through the same classifier.
dsh-checkpoint-diff
tmpdot/dsh-checkpoint-diff
Browses checkpoints produced by dsh-checkpoint-rewind as a timeline of time nodes and shows per-file line diffs between any two of them, with trace replay from session logs (no snapshot producer needed) and preview-first workspace rollback with single-shot undo, via /diff and /rollback commands, a JSON HTTP API, and a GUI panel.
dsh-douyin
anacondakc/dsh-douyin
Short-video sidebar: native player, series navigation, precise history replay.
dsh-feishu-bridge
wz-heng/dsh-feishu-bridge
Fail-closed Feishu (Lark) channel bridge: chat with a bot, get dsh agent turns back — with opt-in human-in-the-loop tool approval (Allow/Deny cards before every bash call, fail-closed timeout). Official-Python-SDK-only integration (exact-pinned); deny-by-default allowlist, webhook signature/timestamp/replay verification, per-chat sticky sessions; daily latest-SDK compatibility canary; bilingual docs.
dsh-selection-explain
yfwu2020/dsh-selection-explain
Select text in the DSH web UI to stream a translation and an explanation of what it means in that specific context, read against the conversation it came from; the panel keeps a replayable history and can be promoted into a full session.
agentduel-dsh
linconz/agentduel-dsh
AgentDuel is a code-battle game. This plugin lets you complete the entire game loop inside DSH: manage deathmatch characters and capture-the-flag teams, submit agent code, start battles, watch replays, and use DSH conversations to keep improving agents based on battle results.
dsh-skillopt
wode25500/dsh-skillopt
Nightly self-improvement for your DeepSeek Harness agent: harvest past sessions offline, replay recurring tasks on your own API budget, and consolidate validated skills behind a held-out gate. Brings microsoft/SkillOpt's SkillOpt-Sleep engine to dsh as native tools plus a bundled skill - no weight training. Requires Python 3.10+ and the SkillOpt-Sleep engine (pip install skillopt).
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-response-id-reuse
gulagala001/dsh-response-id-reuse
Reuses stored Responses IDs for continuation, with retry and branch isolation and verified history replay.
dsh-remote-control (plugin)
moonlitdropofblood/dsh-remote-control
Remote-control DeepSeek Harness from a HarmonyOS phone: list, switch and create sessions; send prompts (the phone follows the host's busy-enter behaviour, queue or steer); stream assistant text, reasoning and tool steps; answer approval requests and questions; view session images as inline thumbnails with on-demand full-size fetch; switch model and reasoning effort; cancel running tasks; receive push notifications. Traffic is end-to-end encrypted (AES-256-GCM with HKDF-SHA256 key derivation) and travels through a dumb-pipe relay that only sees ciphertext; v1.1.3 adds a session-epoch anti-replay handshake and optional native TLS for the relay. The relay can be hosted on the local machine (managed by the plugin) or on a VPS; the phone app and the plugin ship in one repository. 5 Remote methods.
dsh-zcode-plan-bridge
tangyixiao/dsh-zcode-plan-bridge
Requires the ZCode CLI installed and signed in. Relays the ZCode CLI agent into DeepSeek Harness as a model provider. A resident app-server session streams each turn, and the model catalogue is read from the CLI at runtime, so context windows, image support and provider routes follow the installed CLI and account. Stable 0.6.6 supports DSH 0.2.2-rc.1 and 0.2.2-rc.3, adds a bilingual toolkit, optional Z·SWITCH saved-account controls with independent DSH or shared login, next-request quota switching without failed-task replay, and canonical context-overflow handling. Session titles are generated in disposable native sessions. Quota rotation recognizes attributed Start Plan code 1005 with the exact exceed quota limit message. Tool permission requests are denied by default; unknown approval or permission methods are always denied, and full-auto requires an explicit opt-in. DSH approval UI does not mediate this separately spawned app-server process, and the policy is not an OS sandbox. Derived from kyle123740/dsh-zcode-cli-proxy (AGPL-3.0); it registers the same `zcode-cli` provider id, so it cannot be installed alongside the original. Its HTTP routes first require the synchronous connection.requestRejection Host/Origin policy, then connection.admit owner admission; missing, throwing, malformed or asynchronous checks fail closed. Independent rotation continues the current task by default after confirmed quota exhaustion and native tool drain.
dsh-deepsea-echo
cshengyi023-collab/dsh-deepsea-echo
Arknights-flavoured startup animation for the DSH Web GUI: a PRTS terminal boot, sonar scan, hologram lock-on and an operator-file splash that end in a deep-sea lockup, about 4.6 seconds, with four selectable original score presets (or sound off), skippable with a click or Esc and switchable, re-timed or replayed from Settings.
dsh-loop-continue
yunxiyang/dsh-loop-continue
Resumes an agent turn that ended after narrating its next action without calling a tool: the guard replays the turn log on agent/turn-stopping, asks a judge model one true/false question, and steers the same turn to run one more step.
huahua-dsh-record-replay
azure5100/huahua-dsh-record-replay
Full-featured record & replay for DSH (works on Windows) from the huahua plugin series: replay every auto-recorded session as a readable timeline, export/import shareable replay packs, re-run conversations in a fresh session, and record screen operations to auto-generate installable SKILL.md skills. Includes a 4-file Windows adaptation of kangshifu1/dsh-record-replay for DSH rc.5 (see README derivative notice).
dsh-tool-visual-primitives
inkshadewoods/dsh-tool-visual-primitives
Analyzes conversation images through mode-specific prompts (caption, UI, document, grounding, topology, etc.) and injects structured evidence with coordinate primitives (boxes, points, refs) as text, with session-level caching for reuse across replay and compaction.
dsh-session-export
yangyongzhen/dsh-session-export
Export sessions to Markdown/JSON for review, replay, and per-session cost summaries.
dsh-plugin-weaknet-adaptor
jiay98528-dev/dsh-plugin-weaknet-adaptor
Weak-network adaptor: transparent long-backoff model-stream retries (10-min cap), local response replay cache, heartbeat auto-reconnect, and degraded-mode token economy, with a bilingual (zh/en) settings page.
LiveCode
liyu6969/livecode
Right-sidebar panel that replays a file write or edit line by line as the agent produces it, with per-file tabs, a diff gutter and an optional animated character scene.
oh-my-clm
devzjqx/oh-my-clm
Pi extension + DSH plugin: manual live-context cleanup and automatic, replay-safe tool-result clearing.
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.
dsh-super-cards
baitouwuya/dsh-super-cards
Native UI cards instead of Markdown: the assistant calls render_ui and the answer renders as a card in the conversation — 29 block types (metrics, tables, charts, timelines, kanban, flow charts, clickable actions), a content auto-layout mode, four placements (inline, process, dock board above the composer, turn-tail next-step card), a schema-backed settings form, and an optional automatic next-step card. Cards are rebuilt from the recorded tool arguments, so they survive replay and fork; 253 offline tests, no build step.