Plugins

Memoria Markdown LLM-Wiki entre sesiones y persona SOUL.md; por defecto se inyecta tras primer mensaje, solo en la sesión activa. remember/recall/consolidate/forget. Requiere formato v4 (DSH oficial 0.1.7+) y source kind plugin:memory. v0.13 gate engines.dsh exige 0.2.1-rc.5 oficial y rechaza hosts anteriores; Review now: POST /api/memory/control action growth-now envía un resumen que respeta el floor. v0.14 gate por sesión (chip, /memory on|off|status, ajustes): off detiene inyección/digest/avisos/last_access, sin borrar; disabledWorkspaces permite poner proyectos completos en off por defecto; las elecciones de sesión duran 7 días. v0.15 Basic soul escribe un bloque gestionado en SOUL.md sin tocar el resto, completa BOOTSTRAP, refleja la soul existente y avisa si falta. Informe checkup.

Herramienta de shell Git Bash para Windows: ejecuta comandos a través del bash incluido con Git for Windows, con tiempo de espera, sandbox, truncamiento de salida y trabajos en segundo plano, reemplazando la lenta herramienta pwsh y la herramienta bash exclusiva de WSL.

Gestiona servidores MCP, skills y herramientas de tu agente — estado real en contexto (cargado/truncado/expulsado) y conmutadores por sesión.

Plugin de DeepSeek Harness que ahorra tokens de forma agresiva sin afectar la calidad: truncado sin pérdida de cabeza/cola de resultados de herramientas (por defecto), plegado de resultados de grep, delegación a dsh-compaction-basic activada por presión y una subclase CompactionEngine con resumen LLM para el modo agresivo. Ahorra el 97.6% de los tokens de entrada en una suite sintética de 8 escenarios, y todas las reducciones pasan por el flujo de eventos de dsh para que modelo-visible = registrado se mantenga.

Reversión de conversación estilo TRAE: los puntos de control por turno (ventana deslizante de 10 turnos) revierten los archivos tocados y truncan el contexto del modelo in situ, manteniendo el mismo session id. Sólo iniciada por humanos: un comando /rollback o un botón Web por turno (funciona también en turnos interrumpidos), con una vista previa que lista cada archivo afectado y su acción: restaurar, recuperar, eliminar u omitir.

Menciones estilo Codex `@file` para la GUI web: busca y referencia archivos de workspace de todo formato (.java, .vue, PDF, imágenes, sin extensión, ocultos) sin truncado de índice. Reemplaza a dsh-at-file.

Añade edición y borrado a tus propios mensajes. Editar trunca los turnos siguientes y vuelve a enviar el texto nuevo; borrar elimina el mensaje y esos turnos. Cualquiera de las acciones puede revertir los cambios de archivos registrados en los turnos eliminados (se rechaza si un archivo cambió después). La barra lateral conserva una sola conversación.

Insignia de modo preestablecido incrustada en el botón "New Session": haz clic para elegir el preset de agente predeterminado para la siguiente sesión (los nombres largos se truncan con puntos suspensivos para mantener la etiqueta legible).

Continuación automática para la interfaz web de dsh: cuando un modelo local alcanza su límite de output tokens y aparece el aviso max-tokens, el plugin pone en cola un mensaje breve para continuar la respuesta. Detección de truncamiento por dos fuentes, protección contra bucles y sin UI.

Muestra el porcentaje de acierto de caché integrado con tres decimales y amplía la línea de estadísticas del Composer para evitar el truncamiento.

Diagnostica por qué no se reconoce o selecciona una skill (descripción truncada, frontmatter incorrecto o recursos ausentes) y comprueba conflictos entre entradas de cargadores de plugins instalados. Guarda un registro por perfil y compara cada revisión con la anterior, agrupa hallazgos repetidos por causa raíz, crea hojas para reescribir descripciones demasiado largas y enumera los parches de plugins que modifican la configuración del directorio de skills.

Persona y memoria desde un único AGENT.md: se inyectan los primeros 40.000 caracteres en el system prompt de cada sesión; el exceso se trunca con una nota. Se edita desde Settings. Las líneas de diario por turno se fusionan por similitud, se agrupan según la frecuencia de recuperación y solo se depuran al alcanzar el límite de 2 MB.

Recuperación automática ante turnos fallidos: reanuda tras fallo temporal, continúa tras truncamiento de salida y reejecuta si quedó una lista todo incompleta. Incluye respaldo 429 por solicitud y dock para editar y reenviar, reintentar o detener y volver a preguntar.

Rollback de conversación estilo TRAE, reconstruido para DSH 0.1.7-rc.2: checkpoints por turno restauran los archivos modificados y recortan el contexto del modelo in situ bajo el mismo session id. /rollback doctor comprueba el contrato framework-contract; la vista diff de archivos afectados se calcula contra el estado actual en disco para seguir siendo correcta tras reiniciar. Incluye deshacer la última acción con Ctrl+Shift+Z, opción para ocultar mensajes revertidos e interfaz en inglés.

Libera el pie de estadísticas del compositor: las líneas largas de estadísticas de turnos y tokens ocupan todo el ancho del compositor en lugar de truncarse con puntos suspensivos.

Keeps a turn from closing on an unfinished reply: a stream guard turns silent truncation into an in-step retry, and a completion protocol asks the model to declare how each turn ended.

Adds a relay_check tool that runs relaycheck against an LLM API relay to detect model swaps, misreported usage, truncated context and injected prompts; keys are read from environment variables.

Idle-time context truncation for a single DeepSeek Harness session. After a configurable number of idle hours the older conversation is dropped without a summary, so one long-lived session cannot leak one task into the next; it ships a per-session toggle, a tab next to the chat, and a /truncate command that IM bridges can drive.

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.

Títulos automáticos para sesiones DSH: las nombra con el primer prompt y la primera respuesta del asistente (“project｜task”), en vez de truncar el prompt. Sustituye el proveedor de títulos integrado mediante un bundle patch, añade “AI rename” al menú de sesión con diálogo de vista previa y ajustes de proveedor/modelo, reglas y registro de eventos de diagnóstico. 49 unit tests.

Administra servidores MCP, skills y herramientas del agente DeepSeek Harness. Muestra el estado real en contexto (cargado/recortado/expulsado) y controles por sesión para cada capacidad.

Muestra cuánto ocupa cada skill en el contexto del modelo y permite ocultarla del catálogo sin borrarla. Lista todas las skills del disco con la línea exacta del catálogo recibido por el modelo (description normalizada y truncada al límite), tamaño del cuerpo y tokens estimados. Previsualiza byte por byte el mensaje del catálogo tal como lo recibe el agente, lo compara con la copia publicada en la sesión y detecta skills que no llegan al catálogo por frontmatter inválido o shadowing de una raíz prioritaria. Ocultar escribe disable-model-invocation en frontmatter y conserva el archivo, por lo que sigue funcionando mediante `/name`. También importa de ClawHub, Skills.sh y URLs GitHub y desinstala las skills propiedad del usuario.

Archiva tus propias publicaciones de QQ Zone en una biblioteca local, ordenadas en tres columnas —pensamientos, sueños y reseñas (recomendaciones de libros/películas)— corregibles manualmente una por una; deshacer en una no altera las otras. Incluye visor web privado con contraseña, reparación de publicaciones largas truncadas por la API a unos 500 caracteres, recuento por año/mes y copia local de imágenes adjuntas para leer sin conexión. La cookie solo se guarda en esta máquina.