Plugin

Memoria Markdown LLM-Wiki tra sessioni e persona SOUL.md; iniezione predefinita dopo il primo messaggio, solo sessione attiva. Richiede sessione v4 (DeepSeek Harness ufficiale 0.1.7+) e source kind plugin:memory. v0.13 engines.dsh richiede 0.2.1-rc.5 ufficiale e rifiuta host precedenti; Review now POST /api/memory/control action growth-now invia un riepilogo nel rispetto del floor. v0.14 gate per sessione (chip, /memory on|off|status, impostazioni): off ferma iniezione/digest/promemoria/last_access senza cancellare; disabledWorkspaces consente a interi progetti di essere off per default; le scelte di sessione durano 7 giorni. v0.15 Basic soul aggiorna solo il blocco gestito in SOUL.md, lascia intatto il resto e completa BOOTSTRAP. Report checkup con azioni prioritarie.

Motore di contesto Zero-residence: i payload di grandi dimensioni degli strumenti vengono espulsi dal contesto attivo verso un archivio interrogabile e rimangono solo puntatori deterministici, così i token di prompt per turno smettono di crescere con la trascrizione. Qualsiasi contenuto mascherato viene ricostruito alla lettera dal log della sessione su richiesta, tramite id di chiamata dello strumento o seq della sessione.

Consente di approvare chiamate agli strumenti, seguire l’output in streaming e rispondere alle domande dal telefono tramite un relay zero-knowledge, con crittografia end-to-end a livello di payload.

Accedi a DeepSeek Harness con un abbonamento Claude Pro/Max (senza chiave API), riduci le immagini delle richieste Anthropic a un lato lungo di 2000px, alleggerisci i payload del sintetizzatore /compact e mostra l'utilizzo del piano a 5 ore/settimanale più i conteggi dei messaggi di DSH.

Solo Windows. A ogni avvio di DSH controlla senza configurazione se Genshin Impact è installato. Se sì, avvia il gioco; se si chiude subito, ad esempio in attesa di un aggiornamento, apre il launcher miHoYo. Se manca, scarica l’installatore del launcher sul desktop e rifiuta link con un contenuto che differisce di oltre 15 MiB dai ~224 MiB attesi. Se il download fallisce, apre la pagina ufficiale.

Hook del ciclo di vita in stile Claude Code: comandi shell configurati vengono eseguiti prima/dopo le chiamate agli strumenti del modello con un payload JSON su stdin; un codice di uscita diverso da zero nel pre-tool blocca la chiamata

Adds a { } button under each assistant reply in the DSH Web GUI that opens the exact request sent to the model for that reply: model, call parameters, messages, tools, and provider-reported token usage.

Policy-layer guard that checks tool calls before they run: a read whitelist, a sensitive-path deny list (credentials, browser profiles, chat history, other users' home directories), dangerous-command blocking, subagent isolation, and a static scan that quarantines executables. Every denial is appended to an audit log. It is a text-and-path filter, not a sandbox or antivirus: dynamically built paths, encoded payloads and packed binaries can get past it, and it does not replace the OS antivirus.

Precomputes the compaction summary in the background and commits the held result when pressure compaction fires, removing the foreground model round-trip. Registers itself as the session compaction engine, pins the span it summarised so the payload fingerprint still matches at commit time, prunes tool results before committing, and falls back to the stock foreground path on payload drift or context overflow.

Registra database esterni e corpora documentali come strumenti richiamabili dal modello. Include confine di accesso per fonte, ledger dei riferimenti credenziali senza valori e registro fetch con fingerprint e hash dei contenuti, non i payload.

Riprende i turni interrotti dal riavvio dell’Host DSH. Dopo l’attivazione della sessione intercetta agent activation, legge turn/end riparati e invia una continuazione che verifica lo stato reale prima di presumere l’esito di un effetto collaterale. Salta errori permanenti (AUTH, QUOTA, contesto eccessivo) e usa budget tra riavvii, backoff adattivo e payload minimo.

Budget configurabili per turno per le mutazioni di file supportate, che limitano file distinti, chiamate di mutazione e byte di payload UTF-8 prima dell'esecuzione dello strumento.

Cattura ogni payload dell'API del modello upstream in JSON (debug e osservabilità)

Approval frame for DSH tool calls: a rule table with four outcomes (deny, ask, AI review, allow) across tool, command, file and network scope, then AI fast/deep review and human approval, a hash-chained audit trail and a settings UI. Target files are inspected by metadata only (no file contents are opened); the reviewer sees the command, targets and clipped write payloads, with secrets redacted before anything leaves the machine.

Look up Chinese companies inside the conversation: name a company and get its business registration, shareholders, executives, outbound investments and registration changes, plus judicial and operating-risk records such as enforcement, dishonest judgment debtor, consumption restriction, abnormal operation, serious illegality and administrative penalties (11 dimension groups). Results render as conversation cards with fields and the raw payload on expand, and the API key is saved from the plugin page into the Harness credential store.

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.

Plugin di trigger basato su eventi (il campanello accanto alla sveglia di dsh-proactive): fs watcher ricorsivi con finestre debounce e filtri glob per regola; endpoint webhook solo loopback con token per regola; regole che risvegliano una sessione target tramite il servizio agents dell’host (follow-up diretto se attiva, prima ripresa se inattiva) e inseriscono un prompt con variabili paths/count/payload. Ricarica le regole senza riavviare DSH e aggiunge ogni trigger a una cronologia JSONL verificabile, consultabile con trigger_list e trigger_history. Una route health loopback riporta stato installazione/runtime. Esempi: riordinare Download, diagnosticare allarmi CI/monitoraggio da script locali e sorvegliare errori nei log. JavaScript puro, senza build, MIT.

Attiva o disattiva il server rea-mcp dall’intestazione sessione della DSH Web UI o dallo strumento agent rea_mcp_toggle. Con REA spento, i 170 schemi degli strumenti mcp__rea__* non vengono più inviati a ogni richiesta (il payload tools scende da circa 722 KB a circa 95 KB) e il processo figlio REA termina. La scelta è salvata per profilo e ripristinata all’avvio tramite un’espressione disabled a livello di riga.

Trasforma una richiesta in linguaggio naturale in un workbench attivo: un pannello laterale pone alcune domande, il modello crea un artefatto HTML a file singolo e il plugin lo mostra in una finestra ridimensionabile o in una scheda browser indipendente. Le azioni finestra dichiarate sono annotate nel diario di progetto come informazioni, senza payload. I workbench si importano, esportano o installano da un indice di mercato.

Provider proxy che rimodella ogni payload in uscita con una catena di operazioni per categoria prima dell’inoltro. Mantiene monotono il prefisso del prompt tra i turni, evitando di pagare di nuovo per intero la cronologia invariata. I nomi op sono trasporti (exec/http/inline) e accettano algoritmi definiti dall’utente.

Assistente AI per la progettazione di motori: analisi requisiti → identificazione parametri → matrice candidati → verifica dei primi 10. L0 è un filtro ampio a livello di millisecondi (JS puro, nessuna dipendenza da solver), ordina i candidati e prepara il passaggio agli strumenti di precisione L1/L2. Usabile come preset agente locale e plugin Cordis. L1 (RMxprt) e L2 (Motor-CAD) sono riservati in questa versione; il gate rifiuta subito livelli superiori a l0.

Compressore del contesto sicuro per agenti DeepSeek Harness. Preserva semanticamente output essenziali degli strumenti (errori, percorsi, dati JSON e risultati dei comandi) invece di troncare meccanicamente.

Policy preventiva per i risultati degli strumenti DeepSeek Harness: riscrive il testo che corrisponde alle regole regex oppure rimuove/tronca payload imprevedibili in base al percorso del campo, prima di normalizzarli, mostrarli o salvarli. Controparte upstream di dsh-plugin-redact.

Policy preventiva sui risultati degli strumenti: riscrive il testo corrispondente ed elimina o tronca interi campi in base al percorso prima di normalizzare, renderizzare o persistere il risultato, così payload non revisionati non entrano nell’elaborazione. Inerte all’avvio, senza blocklist predefinite di categorie o domini.