メモリ・ナレッジ · DeepSeek Harness プラグイン — 27 ページ

Host-plane memory service for DeepSeek Harness backed by a local OpenViking server: automatic session capture (write-behind with retry + dead-letter), session-scoped recall, and profile injection primitives.

Model-facing memory tool and prompt context

Agent-preset tool package for the host `memory` service: thin memory tools plus an automatic compact user-profile prompt section and per-input targeted injection.

The dsh browser-surface bundle: the web patch layer over dsh-base plus the runtime glue plugin (frontend dist serving, web-surface prompt, bash runtime variables, URL line)

Native, local-first knowledge base for DeepSeek Harness: ingest local documents, retrieve with Chinese n-gram/BM25, and answer with citations

组装式上下文（Assembled Context）: a context tree over the dsh session surface with per-node assemble modes (full / key / off), agent-authored presets, and a browser panel that decides what the model actually sees.

澶?engram 灏?KV锛氬缃?engram 杞帴灞傦紙璺ㄤ細璇濆垎灞傝蹇嗭細global/project/session锛孉I 鑷富鍐崇瓥褰掑睘 + 鍥犳灉閾炬帴绱㈠紩 + 娓愯繘鎶湶 + 涓诲姩鍞ら啋锛夛紝涓?DSH 瀹炵幇瓒呴暱涓婁笅鏂囪蹇嗙█鐤忚矾鐢憋紝涓庡畼鏂?compact 鍏卞瓨锛堝畼鏂硅吘 KV锛宔ngram 淇濈粏鑺傦級

Local-first RAG knowledge tools for DSH: kb_query / kb_ingest / kb_crawl / kb_related, backed by the kb-rag Python pipeline (Ollama bge-m3 + ChromaDB + SiYuan). Zero API cost, data never leaves your machine.

记忆海 · 大肥鱼中枢：把一份本地长期记忆库（JSONL）接成 DeepSeek Harness 的侧边栏面板——看状态、语义检索、调半衰期、跑夜间体检、开星图

/fuck — read your Claude Code & Codex conversation history in parallel, analyze your preferences and tool habits, migrate them into DSH as a persisted profile, and keep remembering them.

DSH 角色设定与长期记忆：角色库（persona.yml + SOUL.md + USER.md）、全局/项目两级记忆（FACT.md / MEMORY.md）、可检索的长期日志 JOURNAL.jsonl（不注入提示词，按需 search）；首条消息快照注入，对所有 agent 预设生效，附右侧栏管理面板与设置页「记忆」注入开关（可扩展的设置项注册表）

dsh-memory — cited memory over DSH's lossless session log: post-session distillation into human-auditable markdown files, a compact recall index, memory_read/memory_expand tools, and a dsh-memory CLI

DeepSeek Harness adapter for transparent memory windows, associative retrieval, and extractive compaction.

Adds a true delete-session capability to the dsh web GUI (today only rename / fork / archive exist): a per-session danger action in the conversation header that removes the session's workspace accounting, archive-set entry, in-memory store row, and its durable event log on disk. Running sessions are refused for safety.

DSH Remote Control 的宿主侧插件：把本机 DeepSeek Harness 与中继配对，微信小程序可发指令、看流式输出、回答审批。载荷级端到端加密。

The injection packet: what a model turn actually sees, projected from facts through dsh-mnemon

ego lite for DeepSeek Harness, with a memory it keeps: the agent recalls what it learned about a site before it acts, promotes a script that worked into a reusable site tool, and hands you the keyboard through a real dsh prompt when a page needs a human.

Troubleshooting knowledge base for ops mode — distilled postmortem cases stored in the access hub, searched before diagnosing and recorded after resolving.

DeepSeek Harness mobile remote-control plugin: LAN/external remote control (QR / on-off toggle / remote-access password gate), rich telemetry (expandable device details, system CPU/memory/load, DSH app state: version/sessions/workspaces/plugins/models), real external-tunnel status detection (frpc token / reconnect diagnostics), mobile-first UI polish (composer image upload entry via system picker, settings single-column layout) and faster remote loading.

Hierarchical Claude Code-style persistent memory for DeepSeek Harness: user + project (git-shared) scopes, session-start auto-injection, MEMORY.md index guards. Pure files, zero dependencies.

dsh plugin: verbatim chapter archive + zero-token index for small-context models

Hybrid memory plugin for DeepSeek Harness: L1 snapshot memory (MEMORY.md/USER.md, read on demand, zero context cost) + L2 searchable knowledge base (one-fact-per-file, SQLite FTS5) + L3 import from Hermes/Claude Code/Codex/WorkBuddy and approved Agent Hub documents. No automatic injection. Data lives on D drive, never C.

Adapt ERP: learn your ERP through the browser and build private local knowledge in DeepSeek Harness

Evidence-backed Kaggle solution research: installable agent skill with offline knowledge