第二篇 · PART 2

Harness The Workbench Around the Model

让模型"长出手脚"的基础设施:工具、文件、沙箱、权限、循环——把只会聊天的模型,变成能真正干活的系统。 The infrastructure that gives the model hands: tools, files, sandbox, permissions, loops — turning a chat-only model into a system that actually works.

什么是 Harness What Is a Harness

Harness(工作台 / 智能体运行框架)是包裹在大模型外围的一层基础设施:它把工具调用、文件与沙箱访问、Shell 执行、上下文管理、权限审批、运行循环与评测组装在一起,让模型不再只是"文字进、文字出",而是可以读取你的文件、运行你的代码、搜索网络、反复试错

A Harness is the infrastructure layer wrapped around an LLM. It wires up tool calling, file & sandbox access, shell execution, context management, permission approvals, the execution loop, and evaluation, so the model can read your files, run your code, search the web, and iterate — not just exchange text.

🧠

模型 = 大脑 Model = Brain

负责思考、理解、生成方案与文本。

Thinks, understands, produces plans and text.

🛠️

Harness = 身体 + 工具箱 + 操作规程 Harness = Body + Toolkit + Operating Rules

负责动手:调工具、读写文件、跑命令,并确保一切在权限与安全边界内进行。

Acts: calls tools, reads/writes files, runs commands — while enforcing permissions and safety.

裸模型 vs 接上 Harness Bare Model vs. Harness-Equipped

能力维度 Dimension 裸模型 Bare LLM 模型 + Harness LLM + Harness
对话 / 生成文本 Chat / generate✅ 唯一能力 only ability
读取本地文件 Read local files✅(沙箱限定范围)(scoped by sandbox)
执行代码 / Shell Run code / shell
访问网络 / 搜索 Web / search
会话持久化与记忆 Session & memory❌(无状态)stateless✅(会话、上下文管理)sessions, context mgmt
反复试错、重跑 Iterate & retry✅(执行循环)execution loop
安全与权限控制 Safety & permissions❌ 无控制 no control✅(审批、资源限制、评测)approvals, limits, eval
一句话总结 One-liner 模型决定"想做什么",Harness 决定"能不能做、怎么做、做完了怎么验证"。
The model decides what to do; the harness decides whether, how, and how to verify it.

Harness 的组成 Anatomy of a Harness

大模型 LLM 思考 · 生成 · 决策 Think · Generate · Decide ← 工具结果回填 Feedback → 文件系统 File I/O读/写/编辑 项目文件read / write / edit Shell / 代码执行运行命令、Python 脚本run commands & scripts Web / 搜索 Search联网检索资料与文档web search & fetch 沙箱与权限 Sandbox限定可访问范围、审批敏感操作scoping & approvals 上下文管理 Context会话记忆、压缩、检索memory, compaction, RAG 循环与评测 Loop & Eval试错重跑、结果检查retry & verify 工具接口 / MCP结构化工具调用、开放协议tool calling, MCP 多智能体编排 Orchestration派生子任务给子代理subagents, fan-out
Harness 组件全景:模型居中,工具与能力节点环绕(点击各页导航可深入)
🧰

工具调用 Tool Calling

模型输出结构化指令("调用 ×× 工具,参数为 …"),Harness 代为执行并把结果回填给模型,模型据此继续。

The model emits a structured call ("invoke tool X with args…"), the harness executes it and feeds results back so the model can continue.

📁

沙箱与文件访问 Sandbox & Files

把模型的"手"限定在指定目录内(如工作区),防止它乱动系统文件;读写都有策略约束。

Scopes the model's hands to a workspace so it cannot touch system files; reads/writes follow a policy.

🖥️

代码执行 Code Execution

让模型直接运行 Python 等命令,看到真实输出与报错,从而"试错—修正"地完成编程任务。

Lets the model run commands, see real output and errors, and fix them iteratively.

🔐

权限与审批 Permissions & Approval

危险或越权操作需征求你的同意;资源消耗可设上限——这是放心让 AI 干活的前提。

Sensitive actions require your consent; resource use can be capped — the precondition for trusting AI with real work.

💬

上下文管理 Context Management

维护会话历史、压缩过期内容、注入必要的工具结果,让模型"记得住"长任务。

Maintains session history, compacts stale content, and injects tool results so long tasks stay coherent.

♻️

循环与评测 Loop & Evaluation

允许模型反复尝试、观察结果、调整方案;任务完成后按目标评测是否真的完成。

Allows repeated attempts, observation, and replanning; evaluates completion against the goal.

MCP:连接工具的"通用插座" MCP: A Universal Socket for Tools

Model Context Protocol(MCP)是 Anthropic 于 2024 年提出的开放标准:它把"模型 ↔ 工具"的对接方式统一起来——工具开发者按 MCP 写一次服务,任何支持 MCP 的 Harness/Agent 都能即插即用,就像 USB-C 一样。

Model Context Protocol (MCP), an open standard introduced by Anthropic in 2024, unifies how models connect to tools: build an MCP server once, and any MCP-compatible harness/agent can plug in — like USB-C for AI tools.

  • 文件系统、数据库、浏览器、GitHub、科学计算环境……都能做成 MCP 工具 Filesystems, databases, browsers, GitHub, scientific computing — all can become MCP tools
  • 越来越多的编程工具(Claude Code、Cursor 等)都支持 MCP More and more coding tools (Claude Code, Cursor, …) support MCP
  • 对射电天文来说,将来可以把 CASA、数据归档查询等封装成 MCP 服务,让 Agent 直接调用 For radio astronomy, one could wrap CASA or archive queries as MCP services for agents to call directly

实例 Spotlight:DeepSeek Harness Case Study: DeepSeek Harness

你现在正在使用的这个网站,就是运行在一个真实 Harness——DeepSeek Harness(DSH)——之上的。看它如何对应上面每个组件:

This very site is powered by a real harness — DeepSeek Harness (DSH). See how it maps to each component above:

组件 Component在 DSH 中的体现 How DSH implements it
工具调用 Tool calling内置 read / write / edit / glob / grep / pwsh(PowerShell 执行)等工具 Built-in tools: read, write, edit, glob, grep, pwsh (PowerShell), …
沙箱与权限 Sandbox & permissions文件操作受工作区策略约束(如 workspace-write),越权操作触发审批 File ops are bound to workspace policy; out-of-scope ops trigger approval
上下文管理 Context会话持久化、工具结果回填、长任务目标追踪(goal tools)Session persistence, result feedback, goal tracking for long tasks
循环与评测 Loop & eval后台任务(background jobs)、完成度评测与阻塞上报 Background jobs, completion/blocker reporting
多智能体编排 Orchestration可派生子代理(subagents)、工作流脚本(workflows)并行分派 Subagents and workflow scripts for parallel delegation
Web / 搜索 Web / search内置 web_search 工具检索最新资料 Built-in web_search for up-to-date info
💡 观察:Harness 的"隐藏价值" The hidden value of a harness 模型负责聪明,Harness 负责可靠。没有 DSH 这样的基础设施,模型再聪明也无法替你建这个网站、无法读你的 FITS 数据——它没有"手"。而 Agent,就是"接上了 Harness、带着目标去完成任务"的那套系统。DSH 的开源理念是 "Everything is a Plugin"(一切皆插件)——能力按需扩展,详见 技能生态篇The model provides intelligence; the harness provides reliability. Without infrastructure like DSH, even the smartest model has no hands. An Agent is precisely the system that — equipped with a harness — pursues a goal. DSH's open-source philosophy is "Everything is a Plugin" — see the Skills & Ecosystem page.

下一步:什么是 Agent? → Next: What is an Agent?