Model-driven context management (Active Context Pruning / ACP) for the DeepSeek Harness — the model decides when and what to compress. Ported from billion-context-pi (ranxianglei); acp-kernel reused verbatim. CompactionEngine backend with compress/decompress/search_context/acp_status tools.

DSH 个人配置:Token 用量统计 / 成本估算 / 自定义指令 / 外观全局字号 / 桌面宠物(含会话状态概览),纯本地不联网。Personal center for DSH: usage stats, cost estimation, custom instructions, global font size & desktop pet with session overview — all local & offline.

Fork of @deepseek-ai/dsh-agent-instructions that delivers the workspace instruction baseline as system-prompt sections, keeping the upstream mid-session reminder path.

Interactive long-session background agents plus persistent multi-agent team rooms for DeepSeek Harness: durable continuable child agents with progress, steering and interruption, and cross-session team rooms with a message bus, shared task board, approval

Active Context Pruning (ACP) for the DeepSeek Harness — model-driven context management as a CompactionEngine backend.

dsh Web UI admin plugin: eleven management panels (extensions, skills, MCP servers, subagents, commands, hooks, automation - cron / webhook / workflow, sessions & web search, usage dashboard, todo dock) inside the official dsh settings UI

Task board plugin for the DSH web GUI: a sidebar entry plus a multi-column kanban view with real execution through DSH sessions, schedules, session rules and auto-cruise automation. Board truth lives on the host and syncs to every device over SSE (the bro

Visual Kanban board for DeepSeek Harness: Gitea and GitHub-backed tasks, workflow columns, and a dedicated agent session per task.

Branch from any answer and keep branching at any depth. Recursively isolated follow-ups that leave the main DSH task untouched.

Bounded, layered, approval-gated, auditable cross-session memory for DeepSeek Harness (capability seam: ctx.memory + SQLite provider + memory tool + frozen snapshot injection)

DeepSeek Harness MCP Connector: connect servers, search tools across active connections, and troubleshoot discovery. Includes a continuously updated catalog; supports OAuth 2.0 PKCE, API keys, Streamable HTTP/stdio, and mcpServers JSON import.

Proactive associative memory for DSH: system-prompt recall before the model speaks, three-layer auto-consolidation, skill crystallization, and Astra-style context management - handoff ledgers, PLAN whiteboard, water-level sensing. Local-first, model-agnostic, zero deps. 主动联想记忆+Astra 式上下文管理：自动唤回/自动沉淀/技能固化/交接账本与白板跨窗口续命/水位感知。

DeepSeek Harness session manager: delete, archive, move across workspaces, migrate presets, favorites, review-later, search, filter, sort, priority, tags, notes, and batch actions. | DSH 会话管理：删除、归档、跨工作区移动、预设迁移、收藏、待回看、搜索、筛选、排序、优先级、标签、备注及批量操作。

DSH 会话管理插件：删除（回收站恢复/彻底清除）、统计、继续/暂停、打开日志目录、对话顶部抽屉、工作区分组与排序、上下文压缩阈值设置。DSH session manager: delete with trash/restore/purge, stats, continue/pause, log folder, header drawer, workspace grouping, context compaction threshold.

Git status badges for DeepSeek Harness: an input-row chip on every install, plus sidebar session-row badges (git status, PR/CI action tokens) after one manual step: `npx dsh-git-badge apply`, then restart dsh web. Event-driven freshness over SSE.

Requires dsh >= 0.1.7 (session format 4); on dsh 0.1.5 use dsh-repeat-tool-breaker@0.7. Escalating repeat detection for dsh: a synchronous ctx.tools.guard gate that counts semantic fingerprints (exact/cmd/net/host/sink) over a per-agent sliding window, wa

Reflect2Evolve memory plugin: layered L1/L2/L3 memory, reflection-weighted value backprop, cross-task policy induction, skill crystallization, and three-tier retrieval for OpenClaw, Hermes Agent, and DeepSeek Harness.

Causal memory layer for AI agents — MCP server that records decision→outcome relationships. Survives compaction.