GitLearnOS

你去学。它收尾。You learn. It closes the loop.

先别看介绍。直接进入一次真实结构的 DeepSeek Harness 学习会话：Agent 已读取演示学习状态，并主动展开今天最值得继续的一项。Skip the pitch. Enter a learning session built around the real DeepSeek Harness structure: the agent has read the sample state and opened today's most useful next step.

脚本演示 · 不调用模型Scripted demo · no model calls

像学生一样点。
Agent 把每一步接住。Learn by clicking.
The agent carries every step forward.

提示词、回答和回执都是固定脱敏数据。它复刻原生面板的打开、收起、下一步引导与学习状态变化，但不会连接模型、读取你的笔记或改动仓库。Prompts, answers, and receipts use fixed synthetic data. It recreates the native panel's open, collapse, next-step guidance, and learning-state changes without contacting a model, reading personal notes, or changing a repository.

high-school-learning 标准模式Standard mode 0 / 6
描述你想学的内容Describe what you want to learn
＋Workspace Write脚本引擎 · 无模型Script engine · no model↑
真实界面结构 · 固定提示词 · 合成学习数据 · 零网络请求Real UI structure · fixed prompts · synthetic learning data · zero network requests

个性化解题 × 动态出题 · Harness Developer PreviewAdaptive tutoring × next-question selection · Harness Developer Preview

根据你刚刚怎么做，
决定接下来怎么教。How you just performed
changes how AI teaches next.

AI 会解题，也会出题。GitLearnOS 真正关心的是：你为什么卡住、现在需要多少帮助、下一道题应该验证什么。每次真实表现都会更新判断；新的证据也可以推翻旧判断。AI can solve and generate questions. GitLearnOS focuses on why you are stuck, how much help you need now, and what the next question should verify. Each real attempt updates the judgment—and new evidence can overturn it.

打开核心版快速开始Open the core-ready Quickstart 查看唯一标准设置指令View the canonical setup prompt
来自学习现场Agent 整理一次 Git 提交 From real learningAgent organizesOne Git commit
课堂笔记Class note二次函数Quadratics
老师反馈Teacher feedback配方法步骤Completing the square
题库错题Practice errors2 道2 questions
自学记录Self-study note图像平移Graph shifts
先确认卡点Diagnose first
只给最小帮助Minimum effective help
这一题决定下一题Each answer shapes the next
判断可修正、可追溯Revisable, traceable state
01

同一句“我不会”，
不该得到同一个下一步。The same “I don't know”
should not lead to the same next step.

week-28-gap-pack.md

表面上：三个人都答错了On the surface: three wrong answers

题目Question
y = −2x² + 4x + 1
学生 ALearner A
不懂系数符号与开口方向Doesn't connect sign to opening
学生 BLearner B
概念懂，但调不出顶点公式Understands the idea; cannot recall the procedure
学生 CLearner C
只是在这一题漏看负号Simply missed the minus sign once
关键Key
先用一个小问题区分原因Use one small question to separate the causes
数学学科库 · 已合并Mathematics · reconciled

诊断后：三条不同的学习路径After diagnosis: three different paths

  • A：先练“符号 → 开口 → 极值”A: practice sign → opening → extremum
  • B：恢复顶点公式，或改用配方法B: recover the formula or complete the square
  • C：确认是偶然误读，不做过度训练C: confirm a one-off misread; avoid overtraining
  • 帮助量：独立尝试 → 最小提示 → 再尝试Help: attempt → minimal hint → retry
  • 下一题由这一次真实表现生成The next question follows from this evidence
7c2a9f1 · safe-auto
02

不是“不会 → 完整答案”，
而是每一步都产生新证据。Not “stuck → full solution,”
but new evidence at every step.

1看真实表现Observe the attempt答案、步骤、错误与已有证据Answer, steps, errors, prior evidence
2确认 blockerLocate the blocker用最小追问区分可能原因Use the smallest diagnostic question
3给最小帮助Give minimal help提示、概念、步骤或材料Hint, concept, procedure, or source
4再独立尝试Try independently again观察在多大帮助下能完成Observe how much support was needed
5变式与复测Variation + recheck不把刚听懂误判成真正会了Do not mistake fluency for mastery
6更新下一步Update what comes next保留证据，也允许推翻旧判断Keep evidence; revise old judgments
真实调度器按时唤起同一个主 Agent 时，到期项目进入When a verified scheduler wakes the same main agent, due items enter due-review，回到第 3 步，只问仍不清楚的内容。 and loop back to step 3 — asking only what is still unclear.

主要成功条件是可观察的进步——通过作答与延迟复查验证，而不是一叠整齐的笔记。原始作答、笔记与外部反馈永远保留；更正以新记录链接，不做静默改写。The primary success condition is observable improvement — verified through answering and delayed rechecking, not a tidy pile of notes. Original answers, notes and external feedback are always preserved; corrections become new linked records, never silent rewrites.

03

一个仓库，分科学科文件夹。
上下文更小，管理更简单。One repository, one folder per subject.
Less context, simpler control.

gitlearnos.yml
dashboard.md
learner-profile.md
subjects/
根目录保存共享配置、总览与跨学科安排；每个学科文件夹只保存本科状态。Root files hold shared configuration and cross-subject planning; each subject folder stays focused.
subjects/math/

数学学科库Mathematics

只包含数学资料、题目与记录。Agent 自动判科和写回，用户无需维护路径。Only mathematics sources, questions, and records. The agent routes and writes back automatically.

goals/sources/knowledge-gaps/reviews/events/

只授权一个仓库；每科独立成文件夹，根目录统一保存配置、总览和跨学科安排。大教材、PDF 与扫描件放在 Project Sources 或授权的本地文件夹，Git 只保存紧凑状态、出处指针与历史。Authorize one repository. Subject folders stay focused while root files hold shared configuration and cross-subject planning. Large textbooks, PDFs and scans live in Project Sources or an authorized local folder — Git stores compact state, provenance pointers, and history.

04

下一轮不用重新猜。
技术留在后台，学习状态属于你。The next session does not start from guesses.
Technology stays backstage; the state is yours.

AGENTS.md

当前学习决策The current learning decision

Agent 只需持续回答三个问题：你卡在哪、现在给什么帮助、下一题应该验证什么。它依据任务表现，不给学习者贴抽象标签。The agent keeps answering three questions: where are you stuck, what help is useful now, and what should the next question verify. It uses task evidence—not abstract labels.

Git = formal memory

Git 保存可修正的状态Git keeps revisable state

  • 会什么、卡在哪、需要多少提示What works, what blocks, help needed
  • 旧判断保留证据，也能被推翻Old judgments keep evidence and can be overturned
local RAG = retrieval

RAG 只负责找材料RAG only retrieves sources

  • 学习状态决定该找什么Learner state decides what to retrieve
  • 材料帮助解题，不替代掌握判断Sources support learning; they do not judge mastery

Git 让下一轮从已确认的位置继续；RAG 根据当前缺口寻找教材；Scheduler 只在合适的时候再唤起一道题。它们都不替学习者作答，也不替真实表现决定“会不会”。Git lets the next session continue from confirmed state; RAG finds sources for the current gap; the scheduler wakes one suitable question at the right time. None of them answers for the learner or decides mastery without real performance.

05

DeepSeek Harness 原生支持 · Developer PreviewNative DeepSeek Harness support · Developer Preview

GitLearnOS
直接运行在 Harness 里。GitLearnOS runs
inside DeepSeek Harness.

这不是一篇兼容说明，也不是把 Skill 复制进项目。安装原生 bundle 后，Harness 会加载 GitLearnOS Host 和会话面板。Agent 能读取学习状态、判断下一步，并用 learning_apply 把一次学习变化写成一笔可撤销的 Git 提交。This is not a compatibility note or a copied Skill. Install the native bundle and Harness loads the GitLearnOS Host and conversation panel. The agent can read learning state, decide what comes next, and apply one learning change as a reversible Git commit through learning_apply.

安装到 DeepSeek HarnessInstall for DeepSeek Harness 它具体支持什么See exactly what it supports
一次写完整One complete write事件、知识缺口、模型、复测和 Dashboard 可以进入同一笔原子提交。Event, gap, model, review, and dashboard changes can land in one atomic commit.
Agent 决定下一步The agent decides what comes nextHost 不替学习者排名；面板只呈现主 Agent 已写入的 Next up 顺序。The Host does not rank the learner; the panel renders the main agent's recorded Next up order.
界面不冒充证据UI is not learning evidence面板提供复习、练习、收尾、问老师和看笔记，正式学习状态仍以 Git 为准。The panel offers review, practice, close, ask-teacher, and notes actions; Git remains the formal learning state.

视觉由用户选择：Vision is the user's choice: 可以直接配置支持图片输入的第三方多模态模型，也可以保留 DeepSeek 主模型并安装视觉/OCR桥接插件。两者都没有启用时，Agent 会提醒用户补充视觉能力，不会猜测图片内容。Configure a third-party multimodal model that accepts images, or keep DeepSeek as the main model and install a vision/OCR bridge plugin. If neither is enabled, the agent asks for a visual capability instead of guessing what an image contains.

从一道真实题开始。
让它成为下一道题的依据。Start with one real question.
Let it shape the next.

打开唯一标准快速开始Open the canonical Quickstart
核心版 starter（完整标准指令见 Quickstart）Core-ready starter (full canonical prompt in Quickstart) 
使用 https://github.com/Guojiz/GitLearnOS 作为模板。
我的私有学习仓库：<本地路径或远程仓库>
先读 QUICKSTART.md，确认我的学习目标、学科和资料，再按其中唯一标准指令部署。
Use https://github.com/Guojiz/GitLearnOS as the template.
My private learning repository: <local path or remote repo>
Read QUICKSTART.md first, then follow its single canonical setup prompt after
confirming my goal, subject, and materials.