dsh-personal-todo：SQLite 持久化的个人待办与 Web 面板
dsh-personal-todo 是 zsspub 维护的 DSH 客户端插件，用 SQLite 持久化待办，提供 Web 右栏面板、对话内交互卡片与 Agent 工具集。待办进度与 Agent 执行状态分离，支持 Session 关联、Host 推送刷新、JSON 导入导出及 @ 引用。适用于 Web profile，安装需固定 GitHub commit 并使用 pnpm dsh plugin --profile web add。
Read Moredsh-memory-vault：DSH 外置 SQLite 长期记忆（修复分支）
xs1023 维护的 dsh-memory-vault 是 hjj588 原版的外置记忆库修复分支，数据存于 ~/.dsh/memory/memory.db，兼容旧库无需迁移。修复对话记录重复、标题被宿主消息污染、长期压缩不可达与预设重复注入四类问题，并新增按工作区归属的项目维度与 scope 检索。安装：dsh plugin --profile web add github:xs1023/dsh-memory-vault，重启生效。支持分层记忆、归档搜索与 cordis.patch.yml 调参。
Read Moredsh-destinywind-memory：DSH 长期记忆库与三层系统提示注入
介绍 rickwindman 维护的 DSH 插件 dsh-destinywind-memory（v2.0.0）：用 memory.sqlite 原样存取记忆，设置页与 HTTP API 增删，全库注入每个会话系统提示，并对硬性约束做运行时 user 快照与末尾核对提醒。涵盖安装 bundle 命令、存储路径与上限、从 v1 Markdown/JSON 迁移注意及推荐写法。
Read More@tr1v3r/dsh-ltm：DSH 本地长期记忆与 CJK 感知检索
@tr1v3r/dsh-ltm 是 tr1v3r 为 DeepSeek Harness 维护的记忆类插件：本地 SQLite 存储，CJK 感知分词与 BM25+n-gram 混合检索，近重复检测、scope 隔离、过期复核与 prompt 召回预算；兼容 memory_write/search/forget 并扩展 update、confirm、list、merge 及 revision CAS。安装命令为 dsh plugin --profile <name> add @tr1v3r/dsh-ltm，默认库路径 $DSH_HOME/memory/ltm.db，不强制联网。本文基于 GitHub README 与 package.json 整理安装、配置与 CLI 用法，并说明从 dsh-memory 迁移与运行权限注意事项。
Read Moredsh-plugin-memory-lite：DSH 本地优先跨会话记忆
dsh-plugin-memory-lite 为 DeepSeek Harness 提供本地优先的跨会话记忆：每轮自动写日志、按关键词注入蒸馏条目，可选空闲或手动整理。数据存于工作区与用户级 SQLite，不联网不上传。本文基于官方 README 介绍三层架构、安装命令、配置项、memory_* 工具与用户级库用法，并说明关键词检索与额度相关限制。
Read MoreBOSS直聘自动投递工具：用 Agent Skill 做有规则的批量投递
BOSS直聘自动投递工具是 SkillHub 社区 Skill（@user_7f2d3808/bosszhipin-auto-apply，v1.0.6，ai-agent），无需 API Key。支持按岗位关键词自动搜索、薪资过滤（示例≥10K）、公司与猎头过滤、30 天去重及 SQLite 持久化。文章说明其定位与核心能力，给出 skillhub CLI 与 Cursor 等客户端的安装路径与命令，并归纳典型对话用法与使用注意。官方页面：https://www.skillhub.cn/skills/user_7f2d3808/bosszhipin-auto-apply
Read Moredsh-personal: A SQLite-Based Personal Data Management and Retrospective Bundle
`dsh-personal` is a Bundle for DeepSeek Harness that stores personal data (experiences, projects, tasks, etc.) in a structured way and supports LLM-assisted summarization. Built on SQLite, it provides 22 tools covering recording, querying, correction, and review. Once installed, it can be operated via natural language, supports Jieba-based Chinese word segmentation for full-text search, and data remains available across restarts. Update and delete operations require a `revision` to ensure data consistency. Note that older database versions cannot be opened directly, and cascading deletes are not supported.
Read Moredsh-session-workbench: Session Library Full-Text Search and View Management
The `dsh-session-workbench` plugin provides comprehensive session retrieval and citation capabilities for the DeepSeek Harness Web Profile. It includes a session library (supporting full-text search, highlighted positioning, and multi-session citation) and a session view (supporting tab bar sorting and hiding). Installation requires configuring a persistent FTS index. The plugin is designed to help developers achieve cross-session context recall and historical traceability.
Read Moredsh-session-search: Cross-Session Content Retrieval
This is a DeepSeek Harness (DSH) web plugin intended to enhance sidebar search capabilities. It enables SQLite FTS5 full-text indexing through configuration overrides and allows filtering sessions by keyword, time range, and content source (user/bot/project name). The plugin has no third-party dependencies and provides persistent settings, matched snippet display, and index pre-warming. To install it, users must modify the profile configuration file; after installation, they can adjust the search scope in real time within the search box, with support for index rebuilding and offline verification.
Read Moredsh-memory-search-plus: DeepSeek Harness Cross-Session Full-Text Search Plugin
`dsh-memory-search-plus` is a standalone web plugin developed specifically for DeepSeek Harness (DSH), designed to address the difficulty of searching through large volumes of historical conversations. The plugin uses SQLite FTS5 to build a local full-text index, supporting cross-session searching, Chinese word segmentation, and message-level location. Users can access it quickly via a sidebar or keyboard shortcuts. The interface style is similar to ChatGPT/VSCode, and the plugin runs entirely locally with no network calls or telemetry, ensuring data security and privacy.
Read MoreDeepSeek Harness SQLite Data Tools Plugin
`dsh-tool-sqlite` is a zero-dependency plugin for DeepSeek Harness (DSH) designed to handle local `.db` files. Leveraging built-in Node.js modules, it provides core features such as listing files, inspecting table structures, and executing read-only SQL queries. To ensure security, the plugin enforces strict restrictions: only read-only operations are allowed, SQL statements are whitelisted, and path boundaries and output row counts are limited. Developers can easily enable it in DSH configuration to explore database data directly within the workspace.
Read MoreUsing dsh-kb-sieve to Build Auditable Local Knowledge Base for DeepSeek Harness
`dsh-kb-sieve` is a memory plugin for DeepSeek Harness maintained by omdsh-dev, which builds local knowledge packages with original text citations and SQLite FTS5 from md/txt/docx/pdf files. This paper verifies the capabilities of the three tools `kb_build`, `kb_query` and `kb_read` by cross-checking with the directory page and GitHub source code, covers the installation commands on the directory page, as well as usage boundaries including the PDF dependency `pdftotext`, incremental building and permission risks.
Read MoreCursor Multi-Agent Swarm Rebuilds SQLite for $1339: Planner/Worker Hierarchy Is the Economics of Agents
In July 2026, Cursor released the Agent Swarm research: relying solely on 835 pages of SQLite documentation, without source code or network access, the multi-agent swarm rebuilt SQLite in Rust and passed all sqllogictest tests. The total cost of the hybrid solution planned by Opus 4.8 plus executed by Composer 2.5 is approximately $1,339, while the cost of the single-model GPT-5.5 solution is about $10,565. This article sorts out the Planner/Worker layered architecture, the new harness coordination mechanism, experimental data and the open-source minisqlite product, and analyzes the cost advantage of the multi-agent approach of "using strong models for planning and weak models for execution" as well as its enlightenment to engineering practice.
Read MoreFastAPI + SQLite: Quickly Build a Lightweight Database API Service
This article introduces the process of quickly building a "Student Information Management" database API service using FastAPI and SQLite. First, dependencies such as FastAPI, Uvicorn, and SQLAlchemy are installed via `pip`. SQLAlchemy's ORM is used to define student data models (including id, name, and age fields) and session management, with Pydantic models for data validation. The core implementation includes CRUD operations (create, read single/all students, update, delete). FastAPI routes bind HTTP methods (POST/GET/PUT/DELETE) to generate student management API endpoints. The database configuration uses SQLite, an embedded database that requires no additional server, with data stored in the `test.db` file. After starting the service with Uvicorn, FastAPI automatically generates Swagger UI documentation for easy testing. This lightweight and easy-to-use solution supports asynchronous operations, making it suitable for small-to-medium projects. Future expansions include multi-table associations or migration to PostgreSQL/MySQL.
Read MoreData Storage Fundamentals: How Python Web Saves User Information with SQLite
This article introduces the basic method of implementing web data storage using SQLite and Flask. SQLite is lightweight and easy to use, built into Python, and requires no additional server, making it suitable for beginners. First, the Flask environment needs to be installed. The core steps include creating a user table (with auto-incrementing id, unique username, password, and email fields), implementing registration (parameterized data insertion) and user list display (querying and returning dictionary results) through Python operations. During operations, attention should be paid to password encryption (to prevent plaintext storage), SQL injection prevention, and proper connection closure. The article demonstrates the data persistence process with sample code, emphasizing that SQLite is suitable for small projects and serves as an entry-level tool for learning data storage, with potential for future expansion of functions such as login authentication and ORM.
Read More