插件

安全的数学表达式求值器，零依赖递归下降解析器。

DSH 插件评测框架：YAML 用例驱动真实 headless agent，断言工具调用/参数/返回与 token 用量，baseline 门禁做 CI 回归。

DSH↔Godot 引擎运行时控制桥：通过游戏内置 TCP 交互服务器启动并操控运行中的 Godot 4.x——8 个工具（场景/UI 检查、GDScript eval、输入模拟、截图、headless 静态操作、脚本编译检查），替代 godot-mcp。

DSH 自动记忆插件:三层记忆(用户级/项目笔记/每日日志)自动注入与检索、每日反思、可视化面板与设置页,支持继承其他 AI 工具的历史记忆。An auto-memory plugin for the DeepSeek Harness Web GUI: three-layer memory (user-level / project notes / daily logs) with automatic injection and retrieval, daily reflections, a visual panel and settings page, and inheritance of memories from other AI tools.

A model-native meta-agent plugin for DeepSeek Harness: the model synthesizes task-specific workflows at runtime from five agentic workflow patterns (prompt chaining, parallel workers, router, supervisor, evaluation loop) and coordinates tools and subagent

Markdown 笔记的双知识图谱 RAG：前置依赖 + 相似关联双图（P/S-Agent 遍历）、段落级匹配、ngram/embedding 混合检索、显式权重反馈闭环，以 MCP 接入。

从 md/txt/docx/pdf 构建可审计知识库包（SQLite FTS5），确定性检索与原文阅读。

Browser automation for DeepSeek Harness: Playwright-powered open/click/type/screenshot/eval tools so the agent can drive real web pages.

Design I/O for coding agents: controllable UI generation via declarations (spec/domain/craft/design/components/template) and contracts (skill/evaluator). Use for product UI—console, dashboard, agent-ops, CJK-first apps.

Agent evaluation platform: benchmark YAML, headless run orchestration, trace-based metrics, and run reports

按需取回薄层：context_query / context_slice / context_grep 三个工具读取已持久化的历史，引用可回放。