Plugins

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.

Consulta prazos de artigos de conferências CCF pelas ferramentas ccf_deadlines, ccf_venue e ccf_update, pelo comando /ccf e por um painel desktop que exporta a lista como calendário. As datas vêm de sincronização ao vivo (CCFDDL é a fonte principal, conferida com uma segunda; diferença de 36 horas ou mais é sinalizada como conflito de fontes). Nenhum dado é incluído; os próximos 30 dias ficam no system prompt.

Manual de julgamento sobre dinâmica estrutural de pavimentos: 59 cartões em 8 capítulos apontam títulos que exageram o conteúdo, fórmulas que falham quando copiadas e números que não devem ser extrapolados nem citados fora de contexto.

Mostra ao lado do composer DSH a cota restante do canal Our Free Model (OFM) atual e lista no painel todos os canais autenticados. Marca canais esgotados; trocar o modelo troca também o canal exibido. Requer dsh-our-free-model com ao menos um canal conectado.

Recibos de entrega por sessão para artefatos locais: compara hashes atuais com uma linha de base registrada, separa alegações do agente da aceitação explícita do usuário e sinaliza arquivos alterados ou indisponíveis. Exporta um recibo Markdown por ferramenta, comando slash ou painel de conversa.

Ferramentas adaptáveis à versão do DeepSeek Harness: lê a versão ativa no gerenciador de plugins e replica as regras de compatibilidade do app-boot (`engines.dsh` primeiro, depois peerDependencies `@deepseek-ai/dsh*`, comparação que inclui prerelease e exceções de versão exata). Avalia todos os plugins instalados e marca estaticamente APIs renomeadas ou obsoletas entre v0.1.6 e v0.2.1. Inclui painel de conversa, ferramenta de agente e rotas HTTP state/scan/report. Somente leitura, offline.

Audita o gasto de contexto por turno: detalha AGENTS.md, catálogo de skills, esquemas de ferramentas e contexto de execução; marca reinjeção de conteúdo inalterado como desperdício; classifica o volume de descrições lidas a cada turno com mais de 60 skills instaladas e a inflação de esquemas por plugin. Determinístico, sem LLM. Quatro ferramentas, incluindo varredura entre sessões.

Compara cada afirmação concluída com a saída de ferramenta lida no mesmo turno. Só se a citação do contraditório aparecer nessa saída, conduz uma correção uma vez no mesmo turno. Sinaliza alegações sem suporte no log e em stderr. Primeiro verifica gratuitamente termos absolutos e números incompatíveis; sem ferramenta no turno, não chega ao modelo.

Auditoria de credenciais pré-publicação: percorre diretórios e detecta API keys, tokens, chaves privadas, strings de conexão DB e senhas fixas antes de commit/publicar/entregar. Sem dependências e offline; achados deixam visíveis apenas os seis primeiros caracteres.

Revisão de texto destinado a pessoas. As ferramentas write/edit exigem a flag human_facing; o plugin revisa em segundo plano conforme regras PROSE.md em camadas, prepara anotações, compara com o arquivo completo e as incorpora via /humanize, /humanize-loop ou post-compaction enquanto a sessão principal continua. /humanize-review processa sob demanda arquivos rastreados pelo git.

Registra `valuation_service`, que lê uma árvore workpaper JSON 01/02/04 e retorna um schema unificado: seis grupos de índices, índices previstos, campos de entrada DCF com marcadores explícitos de ausência, PS/PE comparáveis e IRR. É determinístico e local: dados ausentes de WACC, crescimento, capex, D&A ou capital de giro ficam em branco, sem estimativas; diferenças entre premissas são sinalizadas para revisão.

Registra uma ferramenta `context_extract` no DeepSeek Harness. Comprime documentos longos (relatórios financeiros, contratos, livros contábeis) em Markdown estruturado denso e JSON com métricas principais, tabelas, cláusulas numeradas, entidades e alertas de conferência, para que etapas seguintes não precisem reler o original. Detecta e trata separadamente demonstrações financeiras, contratos legais e livros empresariais; confere balanço e caixa e alerta sobre divergências sem corrigi-las silenciosamente. Somente leitura; nunca altera o arquivo fonte. Requer Python com openpyxl e python-docx.

Estúdio de skins de usuário de War Thunder: seletor de quadrilátero UV por quatro pontos, camadas de arte recortadas ao quadrilátero e exportação PNG/DDS/.blk.

Adiciona ao compositor o botão ✨ e o comando /enhance para estruturar o rascunho, indicar perguntas em aberto abaixo da entrada, seguir o modelo da sessão atual e permitir undo.

Campo minado em uma janela flutuante arrastável, com primeira jogada segura, bandeiras, abertura em acorde e cronômetro.

Memo pessoal por semana ISO: navegação semana/mês/trimestre/ano, arquivo trimestral, análise IA e exportação Markdown. Editar semana passada exige force flag; falhas telemetry da suíte viram relatório de issue GitHub.

Sinaliza pares ICD que exigem atenção (combinações dagger/asterisk e combination codes) conforme GB/T 14396-2016 e notas públicas. Apenas relata os pares; a decisão de codificação cabe ao coder.

Recibos por sessão de artefatos locais: compara hashes atuais com a base registrada, separa declarações do agente da aceitação explícita do usuário, sinaliza arquivos alterados/indisponíveis e exporta recibo Markdown por ferramenta, slash command ou painel.

Extrai do texto o número e a data da fatura, as partes, os IDs fiscais e o trio valor, imposto e total. Verifica se valor mais imposto corresponde ao total e sinaliza documentos divergentes.

Verificador de compatibilidade de plugins dsh: compara peerDependencies dos plugins instalados com a versão dsh em execução para apontar os que o preflight de inicialização recusará e exibe comandos de correção copiáveis. Um banner no topo do dsh web permite isolar, desinstalar ou atualizar; faz backup antes da mudança e reverte se a nova checagem falhar.

Selo do saldo da DeepSeek API no rodapé da barra lateral, ao lado de Settings: lê a chave já armazenada pelo DSH, mostra ao passar o cursor a divisão entre crédito concedido e recarregado, alerta abaixo de ¥2 e atualiza a cada 60 segundos ou ao clicar.

Fixa sessões em uma única seção Pinned acima de todos os grupos de workspace na barra Web, sem alterar a ordem armazenada; o pin é um sinalizador no documento de configurações do host. Menu de contexto permite renomear, fixar, bifurcar, arquivar, excluir e copiar título ou ID da sessão, com rótulos no idioma da GUI.

Integração Git para DSH inspirada no VCS do IDEA: o ícone Git junto ao modo de permissão mostra branch e alterações pendentes; ao passar o cursor abre o seletor de branch e ao clicar o painel VCS estilo IDEA (árvore, grafo de commits, alterações e staging, detalhes e diff de arquivo). Inclui ajustes de identidade do commit, binário Git e flags de rede. Não registra ferramentas de modelo no DSH.

QQ Mail e qualquer caixa IMAP/SMTP no DeepSeek Harness: as ferramentas Agent `qqmail_*` listam, pesquisam, leem, enviam, respondem, sinalizam e baixam e-mails via IMAP/SMTP com código de autorização QQ Mail (sem OAuth nem redirecionamento 2FA). Inclui CLI `qqmail`.