Plugins

Audio tools for DSH: inspect audio/video files, transcode or extract the audio track from a video, trim and normalize, and transcribe speech with local whisper.cpp or an OpenAI-compatible API. Requires ffmpeg and ffprobe on PATH (not bundled); local transcription also needs a whisper-cli binary and a ggml model.

DeepSeek account-backed web search and page fetching for DeepSeek Harness sessions whose chat model runs through a third-party subscription or gateway (for example OpenCode Go): registers deepseek-account-search and deepseek-account-fetch providers that call the official Search and Fetch APIs with the signed-in DeepSeek account token, so the built-in web_search and web_fetch keep working without a DEEPSEEK_API_KEY.

IMAP mail triage and cleanup for DeepSeek Harness: incremental read-only scan, a hard-protection rule layer plus optional local-model classification, a Markdown report and a hashed manifest, an explicit approval card, and only then a move to the Trash mailbox — no expunge.

Picks PTC or native tool presentation per task and switches it at runtime for the model or the user.

Bilibili companion for DSH (MIT fork of Alterssss/dsh-bili-whale; do not install alongside the upstream — both register the same `/dsh-bili/` settings routes and share `~/.dsh/bili-whale.config.json`): wakes on a schedule, reads videos through subtitles, danmaku, hot comments and covers, writes two diaries, and shares at most one pick per wake by commenting on the video and @-mentioning the owner (daily quota 3; DM when `share.via=dm`). Inbound DMs, commenter names, summary requests and live danmaku are normalized and scanned for injected instructions before they reach the model; only the owner DM channel can carry an authorization prefix.

Three ledgers for a model that loses context between restarts. Self-note: what I have become — capped at 40 entries, and every eviction first leaves a one-line crease recording which entry was dropped and why. Judgment ledger: record a claim together with how to verify it, come back when it is due, and mark it right or wrong; wrong ones stay in a mistakes book so you stop assuming you were always correct. Handoff: where this run stopped and what the next one should read first — a baton, not a memory. Three tools each taking an action parameter instead of a dozen separate ones (the prompt is paid every turn). Data lives in ~/.dsh/*.jsonl and is shared with the same-named CLI scripts. Zero dependencies.

Local three-band memory for DeepSeek Harness: a 24h protection band, query recall and scene replay, injected into sessions without embedding or model calls.

Elides stale, oversized plain-text tool results from the model-visible session surface once the prompt cache goes cold, replacing each with a stub that keeps the tail lines and any error lines, and reports per-request and cumulative savings in the composer footer.

Delivers a queued chat message at the next step boundary instead of waiting for the whole turn to end, without aborting the model call already in flight; aborting is an opt-in mode rather than the default.

Replaces the built-in compaction provider: at a configurable fraction of the context threshold (85% by default) a detached task summarizes everything up to the previous turn, and the next step boundary commits that held summary as a checkpoint with zero model calls, so the swap itself takes milliseconds. /bcompact \[message] starts the same generation on demand and then sends the message. A settings page sets the trigger ratio, the summarization model, and how many consecutive failures stop background generation until a half-open probe gets through. Subagent sessions, a changed prefix or an expired summary fall back to the built-in synchronous path.

Cross-session reference cards: drag a Q&A card out of any session into another one and it stays there; when you ask a question in the target session, the dragged cards' question, the answer it got at the time and its attachments are injected into the prompt as earlier conversation context, so the model can quote them directly.

Local Ollama worker tools for DeepSeek Harness: model inventory, a single unverified ask, and a batch runner that chunks long text, checks every item deterministically, returns a small verbatim sample for review and can optionally write the labelled result inside the session workspace.

Named model routes for subagent delegations, configured from a settings page: each alias binds a subagent provider, model and optional reasoning effort, token cap, agent preset, tags, plus a "when to use" line the model reads to pick it, so a delegation names only the alias. Adds five tools (delegate, list routes, edit config, wait, release), shows the route a child actually ran on in its session header, lets each UI element be switched off, and keeps all configuration in one local JSON file instead of YAML.

Manages the lifecycle of the local UM GenAI proxy (127.0.0.1:8788) that DSH's um-genai provider talks to: wakes it on demand for llm/stream calls, stops it after an idle period, and reports health plus a read-only status endpoint. Also ships a DSH settings page for port, always-on mode, idle minutes, interpreter and paths, plus the UM account and proxy API key, and a button that writes the providers.um-genai entry into the profile patch and fills in its model list, since a fresh profile has no provider and the model picker would otherwise stay empty. A renewal watchdog logs in to the campus gateway on a hidden Windows desktop when the upstream token expires. The proxy itself is um_opencode_setup.py, a single-file tool distributed by the University of Macau; a copy is vendored in vendor/ so a fresh install works out of the box, while a copy in ~/.um-opencode/ takes precedence when present. The plugin still does not produce tokens: those come from logging in to the campus gateway. Requires Windows (process discovery and idle shutdown use netstat/taskkill; renewal uses PowerShell and a hidden desktop), Python 3.10+ and a University of Macau account.

Connect ChatGPT subscription models to DSH via OAuth, preserving local tool-call identity across conversation replay.

Adds an "Add model" entry to the model menu for third-party providers, plus a management dialog that switches the model of the current conversation, edits or removes entries, and highlights the one in use.

Fills in a cheaper sibling route for subagent calls that name no provider, model or effort (for example Opus to Sonnet 5.5 at medium, Qwen max to Qwen flash); explicit choices and subagent_fork are left alone. Requires the subagent-model-selection policy to be enabled with the target route in its allowedModels - without it the plugin does nothing.

Caps every image sent to a model at a per-side pixel limit (default 2000 px) by wrapping attachments.readImageRequest, which fixes Anthropic many-image request errors without touching stored images.

Connects Qoder models to DeepSeek Harness through the official Qoder CLI over native ACP stdio (no reverse-engineered model API), with qoderclicn (China) having a verified real ACP tool loop; qodercli (international) is selectable but its real inference path is not yet verified. Requires Node.js 22.18+ and the chosen Qoder CLI installed and logged in on the same machine; the plugin does not install or manage accounts.

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.

Adds a Speed setting (Standard, Fast, Ultrafast) for OpenAI and Codex models on a patched dsh build: the pinned dsh 0.2.1-alpha.2 web client needs the Model / Effort child-slot patch, the host pi-ai must be 0.87.1 or 1.1.0 (other versions refuse to load), and the terminal controls need a patched dsh-TUI. Then it appears in the Model / Effort menu; in dst /speed opens a picker or sets the speed directly, and the current speed shows in the bottom bar. The choice is kept per session, provider and model and is sent as the request service tier.

Signs in to LightVela by WeChat QR code, reads the model gateway key from your own instance and writes the gateway into llm-pi-ai so its models appear in the picker.

Talks to the two public Hugging Face Spaces (MiniMax-VL-01 and Llama-2-7B-Chat) that grok3ai.org only embeds in an iframe, presenting them as OpenAI-compatible models.

Runs a local OpenAI-compatible relay that forwards DSH chat requests to grok.online, exposing its two chat models in the model picker without an API key.