From ae74e4690c803815c24dc0f8f4270e79de1fdbb3 Mon Sep 17 00:00:00 2001 From: dyuan311 <98677639+dyuan311@users.noreply.github.com> Date: Thu, 17 Sep 2026 00:02:30 +0800 Subject: [PATCH 1/2] Publish bilingual usage handbook and homepage announcement. Add handbook pages, document Skill install templates, and announce the release on the Updates section. Co-authored-by: Cursor --- .gitignore | 2 + handbook/HANDBOOK.md | 313 +++++++++++++++++++++++++++++++++++ handbook/HANDBOOK.zh-CN.md | 329 +++++++++++++++++++++++++++++++++++++ index.md | 30 ++++ tests/test_updates.py | 18 +- 5 files changed, 688 insertions(+), 4 deletions(-) create mode 100644 handbook/HANDBOOK.md create mode 100644 handbook/HANDBOOK.zh-CN.md diff --git a/.gitignore b/.gitignore index e458ed5..8e981cf 100644 --- a/.gitignore +++ b/.gitignore @@ -1 +1,3 @@ .worktrees/ +_site/ +tests/__pycache__/ diff --git a/handbook/HANDBOOK.md b/handbook/HANDBOOK.md new file mode 100644 index 0000000..01e74a2 --- /dev/null +++ b/handbook/HANDBOOK.md @@ -0,0 +1,313 @@ +--- +layout: default +title: VeryMath Handbook +permalink: /handbook/HANDBOOK.html +--- + +# VeryMath Handbook + +[中文说明](/handbook/HANDBOOK.zh-CN.html) + +VeryMath is a collection of mathematical tooling for coding agents, in two main parts: + +- **Skill packages**: install them into your coding agent, and the agent invokes them automatically or when you name them. +- **Standalone programs, such as agent harnesses**: tools you deploy and run yourself, with their own UI or CLI. + +--- + +## 1. Installing a Skill + +Every AI4Math Skill installs the same way: **open your coding agent and send it a message asking it to do the install**. You do not run commands or configure paths yourself. + +This works with Codex, Claude Code, OpenCode, Cursor, or any coding agent that can read and write local files and run terminal commands. + +### General template + +Send this to your agent, substituting the repository and the packages you want: + +```text +Please install these AI4Math Skills for me. + +Repository: +Branch: main +Skill paths: +- skills/ +- skills/ + +Steps: +1. Clone or update the repository locally. +2. Read README.md, SKILL.md, and each target Skill entrypoint. +3. If this environment supports local Skill discovery, link each directory that contains SKILL.md into the local skills directory. +4. Keep shared sibling support directories in place when a Skill depends on them. +5. Verify that the installed Skills are discoverable. +6. Tell me the installed paths, whether a restart is needed, and give me one test prompt. +``` + +The agent figures out where your environment keeps its skills directory (`~/.codex/skills` for Codex, `~/.claude/skills` for Claude Code, `~/.config/opencode/skills` for OpenCode, elsewhere for others). You do not need to know. + +### Worked example: `math-paper-reading` + +This walkthrough uses [`math-paper-reading`](https://github.com/VeryMath/AI4Math-Paper-Reading) from the AI4Math · Paper Reading repository (`skills/math-paper-reading/`). + +**You:** + +```text +Please install the AI4Math math-paper-reading Skill for me. + +Repository: https://github.com/VeryMath/AI4Math-Paper-Reading +Branch: main +Skill path: skills/math-paper-reading + +Steps: +1. Clone or update the repository locally. +2. Read README.md, SKILL.md, and the entry files under skills/math-paper-reading/. +3. If this environment supports local Skill discovery, install the directory that contains SKILL.md into the local skills directory. +4. Keep shared sibling support directories in place if this Skill depends on them. +5. Verify that the installed Skill is discoverable. +6. Tell me the installed path, whether a restart is needed, and give me one test prompt. +``` + +**The agent:** clones the repository, reads `skills/math-paper-reading/SKILL.md`, installs that directory into your local skills directory, and reports the path. + +**Verify install:** in a fresh session with a prompt that names the skill: + +```text +Using math-paper-reading, what is your first workflow step when given a PDF math paper? +Do not read a file yet; outline the process exactly as the Skill specifies. +``` + +If the answer clearly reflects the installed `SKILL.md` (steps and terminology match the repo docs) rather than a generic summary, the install worked. + +In daily use you do not need to name a Skill every time; the agent matches the relevant installed Skill from your task on its own. + +### Updating and removing + +Also one message: + +```text +Update the AI4Math Skills installed on this machine: pull the latest main branch +for each repository, sync the installed Skill directories, and tell me what changed. +``` + +```text +Uninstall the Skill: remove it from the local skills directory +but keep the cloned repository. +``` + +### Things to know + +- **A Skill is instructions, not a runtime.** The SageMath Skill needs SageMath on your machine, the Lean Skills need Lean 4 and mathlib, the optimization Skills need their solvers. The Skills themselves usually guide the agent through installing these, but they ask for your approval first. +- **Review anything that downloads, installs an environment, or creates a conda environment before approving it.** +- **Account logins are yours to do.** Never hand an agent an API key to commit into a repository. +- A restart may be needed before a newly installed Skill is discovered; the agent will tell you. + +--- + +## 2. Applications + +### VeryMath Textbook Copilot + +A self-hosted course workspace. Read your textbook PDF on the left, talk to the Copilot on the right. Import your own textbook, connect a coding agent, and you can ask about any chapter, get explanations, generate exercises, mind maps, knowledge graphs, and LaTeX slides. All data stays on your own machine (`~/.course-copilot/` by default) and runs under your own agent account. + +**Core features** + +- Textbook parsing: turn selected pages or chapters into a readable digital textbook with body text, formulas, images, and outline +- Explanations: layered treatment of concepts, theorems, algorithms, and worked examples, with figures where useful +- Quiz generation: exercise cards with progressive hints and reference answers +- Mind maps and knowledge graphs: hierarchical knowledge trees and concept networks, with nodes that jump back to the source page +- Slides: compile chapters into 16:9 PDF decks via LaTeX Beamer +- Free-form Q&A: no fixed template, organized around whatever you ask +- Supporting materials: attach handouts, solutions, and references per textbook, with full-text search and OCR for scans + +**Requirements:** Node.js ≥ 22.13. Slide generation needs XeLaTeX, Beamer, ctex, and CJK fonts. OCR needs Tesseract (plus `chi_sim`/`chi_tra` for Chinese) and Poppler. On Windows, install everything inside WSL2. + +**Installation** is again a single message to your agent: + +```text +Pull https://github.com/VeryMath/VeryMath-textbook-copilot and follow +skills/verymath-install/SKILL.md to deploy VeryMath Textbook Copilot on this machine: +check and install the required dependencies, configure the built-in course Skills, +connect my existing agent, start the workspace, and give me the URL. +``` + +The agent installs dependencies, configures the built-in course Skills, connects your agent, starts the service, and returns the address (production defaults to `http://127.0.0.1:4173`, bound to localhost only). In the browser, click "Import textbook" to select a PDF, then connect your agent under workspace settings. + +The bundled course Skills are `textbook-parse`, `explain`, `quiz`, `mindmap`, `knowledge-graph`, and `textbook-to-ppt`, plus `verymath-install` which performs the deployment. + +Supported agents: Codex, Claude Code, OpenCode, Cursor, Gemini CLI, Copilot CLI, Qwen Code, Kimi Code, Kiro CLI. Pick one. + +Current releases: v0.1.0 (macOS preview), v0.1.1 (Windows preview). + +Full documentation → [VeryMath-textbook-copilot](https://github.com/VeryMath/VeryMath-textbook-copilot) ｜ [User guide](https://github.com/VeryMath/VeryMath-textbook-copilot/blob/main/docs/user-guide.md) ｜ [Architecture](https://github.com/VeryMath/VeryMath-textbook-copilot/blob/main/docs/architecture.md) ｜ [Skill development](https://github.com/VeryMath/VeryMath-textbook-copilot/blob/main/docs/skill-development.md) + +--- + +### Co-Mathematician + +Turns a mathematical research conversation into a file-backed project. The formula is "coding agent + repo filesystem + gates + reviewer loop = research workspace": the agent's main thread acts as Project Coordinator, the research question, goals, status, and messages live in `workspace/project/`, proof, computation, literature, and review work each become their own workstream, and the final paper is rendered only from reviewed reports. + +The Python harness does not run agents. It initializes files, appends messages, creates approved workstreams, checks gates, and renders the working paper. Your coding agent does the work. + +**How it works** + +- A goal must be marked `approved` before it can receive workstreams; draft goals are not executable +- Workstream kinds are `proof`, `computation`, `literature`, and `review` +- Each report carries provenance for important claims, explicit uncertainty, failed explorations, and independent reviewer output under `reviews/` +- `workspace/final/working_paper.md` is a working paper, not a chat summary +- AI4Math domain Skills can be installed into a project's `.agents/skills/`, and the Coordinator suggests and hands off to them as needed + +**Installation**, again through your agent: + +```text +Install Co-Math from https://github.com/VeryMath/co-mathematician.git, +set ~/CoMathProjects as the projects directory, and create a project named +Muon Convergence. Return its path but do not start the research yet. +``` + +Then open that project directory in your coding agent and say "Continue this Co-Math project." Day to day, manage projects with `co-math list / resume / next / archive / reopen`. + +Full documentation → [co-mathematician](https://github.com/VeryMath/co-mathematician) + +--- + +## 3. The Skill repositories + +### AI4Math · SageMath — exact symbolic computation + +Verified SageMath computation for mathematical agents: the agent searches the bundled SageMath reference first, writes Sage code in ordinary Python form, executes it locally, and reports results backed by real runtime output. Covers algebra, number theory, combinatorics, graph theory, polynomial rings, matrices, calculus, finite fields, elliptic curves, Galois groups, coding theory, cryptography, manifolds, and modular forms. Windows users should install and run SageMath inside WSL. + +| Package | Use it for | +| --- | --- | +| `sagemath-skill` | Reference lookup, code execution, and reproducible results | + +Full documentation → [AI4Math-Sagemath-skill](https://github.com/VeryMath/AI4Math-Sagemath-skill) + +--- + +### AI4Math · Auto Research — problem discovery and research loops + +Turns broad mathematical intent into structured problems, bounded research runs, and reviewed proof plans. The largest repository by package count; it also includes helpers that install and drive external research systems (Rethlas, Archon, Danus). + +| Package | Use it for | +| --- | --- | +| `discover-math-problems` | Ranked problems, conjecture lattices, evidence ledgers, and work orders from fuzzy background | +| `proof-blueprint-review` | Proof blueprints, verifier-style reviews, repair hints, strict proof-status reports | +| `open-problem-research-pipeline` | Staged literature search, method analysis, proof framework, and report workflow | +| `graph-theory-researcher` | Read graph theory papers, select questions through user gates, produce verified LaTeX/PDF | +| `agent-laboratory-workflow` | Deploy, validate, and launch bounded Agent Laboratory runs | +| `rethlas-helper` | Install Rethlas and run natural-language proof generation from any project | +| `archon-helper` | Install Archon and run formalization workflows | +| `danus-helper` / `danus-helper-dsh` | Install Danus (OpenCode / DeepSeek Harness variants) for multi-agent research, monitoring, summaries, and paper writing | + +Full documentation → [AI4Math-Auto-Research](https://github.com/VeryMath/AI4Math-Auto-Research) + +--- + +### AI4Math · Paper Reading — deep reading and extraction + +Structured reading of mathematical papers, theorem dependency extraction, and conversion of papers into reusable Skills. + +| Package | Use it for | +| --- | --- | +| `math-paper-reading` | Deep reading, theorem dependencies, proof pathways, local reference notes | +| `paper-to-skill` | PDF preparation, paper triage, proof-pattern extraction, cross-paper synthesis | +| `after-ocr` | Audit and repair formula-heavy OCR Markdown with coverage tracking, evidence logs, multi-pass reconciliation | +| `graph-theory-paper-reading` | Deep-read graph theory papers into structured LaTeX reports with proof trees | + +Full documentation → [AI4Math-Paper-Reading](https://github.com/VeryMath/AI4Math-Paper-Reading) + +--- + +### AI4Math · Writing — papers and slides + +Writing from verified material. It does not replace source evidence, proof checking, or experiment validation; every claim should trace back to a source. + +| Package | Use it for | +| --- | --- | +| `paper-writing` | Draft, revise, structure, and audit papers, with claim-evidence review, proof-obligation checks, notation checks, formula readability, and LaTeX preflight | +| `math-beamer` | Source-grounded Beamer decks: auditable templates, compilation, layout inspection, slide-to-source ledger | + +Full documentation → [AI4Math-Writing](https://github.com/VeryMath/AI4Math-Writing) + +--- + +### AI4Math · Lean Agents — Lean 4 formalization + +A structured way for agents to inspect Lean projects, set up reusable Lean/mathlib workspaces, formalize theorem statements, repair proofs, complete `sorry`s, and validate patches locally. Final patches must not introduce `sorry`, `admit`, new axioms, or silent theorem-statement drift. + +| Package | Use it for | +| --- | --- | +| `lean-setup` | Install or verify Lean 4, `elan`, `lake`, and mathlib workspace readiness | +| `lean-formalization` | Formalize statements, repair proofs, complete `sorry`s, review patches, coordinate optional backends | + +`skills/lean-runtime/` is a shared support layer (scripts, schemas, prompts, tests) that you never invoke directly, but it must stay next to the two public packages. Numina, Archon, and other backends are optional adapters requiring your explicit approval. + +Full documentation → [AI4Math-Lean-Agents](https://github.com/VeryMath/AI4Math-Lean-Agents) + +--- + +### AI4Math · Optimization — modeling and solving + +Modeling mathematical programs, selecting and configuring solvers, and reporting evidence from runs. + +| Package | Use it for | +| --- | --- | +| `optskills` | 103 released OptSkills problem-archetype cards, usable on their own for modeling and solving | +| `mixed-integer-programming` | MILP/MIP with binary, integer, and continuous decision variables | +| `second-order-cone-programming` | SOCP modeling and cvxpy-based conic solver workflows | +| `osqp-solver` | OSQP modeling, repeated solves, status gates, independent verification for continuous convex QPs | +| `cdopt-optimization` | CDOpt and manifold-constrained optimization: modeling, validation, runner generation, evidence reports | +| `or-solver` | Dependencies, installation, licenses, environment variables, and setup troubleshooting for a chosen solver | + +The division of labor: you or the modeling Skill choose the solver; `or-solver` configures that solver. + +Full documentation → [AI4Math-Optimization](https://github.com/VeryMath/AI4Math-Optimization) + +--- + +### AI4Math · Computational Mathematics + +Turning mathematical objects, equations, data, or paper excerpts into reviewed computational representations and reproducible evidence. + +| Package | Use it for | +| --- | --- | +| `invariant-computation` | Route and validate algebraic, topological, geometric, TDA, and certified numerical invariants | +| `least-squares` | Linear, polynomial, nonlinear, regularized, constrained, and Bayesian least-squares fitting | +| `scientific-computing-reproduction` | Reproduce, diagnose, tune, visualize, and report research code with human approval checkpoints | + +Full documentation → [AI4Math-Computational-Mathematics](https://github.com/VeryMath/AI4Math-Computational-Mathematics) + +--- + +### AI4Math · Evolving — iterative experiments + +Currently focused on bounded OpenEvolve experiment sessions. + +| Package | Use it for | +| --- | --- | +| `openevolve-experiment-workflow` | Inspect or create OpenEvolve projects, validate runtime config, run bounded probes, summarize metrics, guide iteration | + +Full documentation → [AI4Math-Evolving](https://github.com/VeryMath/AI4Math-Evolving) + +--- + +### AI4Math · MathTool — standalone tools + +Standalone mathematical tools and lightweight agent adapters. Each package keeps its own workflow, dependencies, provenance, and tests. + +| Package | Use it for | Status | +| --- | --- | --- | +| `math-glossary` | Build, review, back up, import, export, and maintain bilingual mathematical glossaries | Released as v0.1.0 (MIT) | +| `matlab-runner` | Route explicit MATLAB tasks through a MATLAB MCP server with execution, testing, and numerical-validation evidence | Review candidate, unreleased | + +Full documentation → [AI4Math-MathTool](https://github.com/VeryMath/AI4Math-MathTool) + +--- + +## 4. Website + +The VeryMath project homepage: organization overview, repository navigation, and updates. + +- [VeryMath homepage](https://verymath.github.io/) +- [中文使用手册](/handbook/HANDBOOK.zh-CN.html) · [English Handbook](/handbook/HANDBOOK.html) +- [Site source](https://github.com/VeryMath/verymath.github.io) diff --git a/handbook/HANDBOOK.zh-CN.md b/handbook/HANDBOOK.zh-CN.md new file mode 100644 index 0000000..8488a5c --- /dev/null +++ b/handbook/HANDBOOK.zh-CN.md @@ -0,0 +1,329 @@ +--- +layout: default +title: VeryMath 使用手册 +permalink: /handbook/HANDBOOK.zh-CN.html +--- + +# VeryMath 使用手册 + +[English](/handbook/HANDBOOK.html) + +VeryMath 是一套给 Coding Agent 用的数学工具集合，主要包括下面两类： + +- **Skill 包**：装进你的 Coding Agent 后，agent 在遇到对应任务时会自动或按你点名调用。 +- **独立程序，例如 agent 的 harness**：独立部署运行的工具，自带界面或命令行。 + +--- + +## 一、如何安装一个 Skill + +所有 AI4Math Skill 的安装方式都一样：**打开你的 Coding Agent，把一段话发给它，让它自己完成克隆、安装和验证**。不需要你手动敲命令或配路径。 + +支持的 agent 包括 Codex、Claude Code、OpenCode、Cursor 等任何能读写本地文件、执行终端命令的 Coding Agent。 + +### 通用模板 + +把下面这段发给 agent，按需替换仓库地址和要装的包： + +```text +请帮我安装这些 AI4Math Skill。 + +仓库：<仓库地址> +分支：main +要安装的 Skill 路径： +- skills/<包名1> +- skills/<包名2> + +步骤： +1. 把仓库克隆或更新到本地。 +2. 读 README.md、SKILL.md，以及每个目标 Skill 的入口文件。 +3. 如果当前环境支持本地 Skill 发现，把每个含 SKILL.md 的目录装到本环境的 skills 目录。 +4. 如果某个 Skill 依赖同级的支撑目录，保持它们在原位。 +5. 验证安装后的 Skill 能被发现。 +6. 告诉我安装路径、是否需要重启，并给我一条测试指令。 +``` + +Agent 会自己判断当前环境的 skills 目录该放在哪里（Codex 是 `~/.codex/skills`，Claude Code 是 `~/.claude/skills`，OpenCode 是 `~/.config/opencode/skills`，其他 agent 各有配置），你不用关心。 + +### 示范：安装 `math-paper-reading` + +下面以 [AI4Math · Paper Reading](https://github.com/VeryMath/AI4Math-Paper-Reading) 仓库里的 `math-paper-reading` 为例（Skill 在 `skills/math-paper-reading/`）： + +**你说：** + +```text +请帮我安装 AI4Math 的 math-paper-reading Skill。 + +仓库：https://github.com/VeryMath/AI4Math-Paper-Reading +分支：main +Skill 路径：skills/math-paper-reading + +步骤： +1. 把仓库克隆或更新到本地。 +2. 读 README.md、SKILL.md，以及 skills/math-paper-reading/ 的入口文件。 +3. 如果当前环境支持本地 Skill 发现，把含 SKILL.md 的该目录装到本环境的 skills 目录。 +4. 如果该 Skill 依赖同级的支撑目录，保持它们在原位。 +5. 验证安装后的 Skill 能被发现。 +6. 告诉我安装路径、是否需要重启，并给我一条测试指令。 +``` + +**Agent 会做：** 克隆仓库 → 读 `skills/math-paper-reading/SKILL.md` → 把该目录装进本机 skills 目录 → 回报路径。 + +**验证安装：** 新开一个会话，发一条点名调用的指令： + +```text +用 math-paper-reading 说明：接到一篇 PDF 数学论文时，你的第一步工作流是什么？ +先不要真的去读文件，只按 Skill 里的规定概述流程。 +``` + +如果 agent 的回答明显来自刚安装的 `SKILL.md`（步骤、术语和仓库文档一致），而不是泛泛而谈，说明装好了。 + +当然，日常使用时不必每次点名 Skill；agent 会根据你的任务自行匹配已安装的相关 Skill。 + +### 更新和卸载 + +同样是一句话： + +```text +更新我本机安装的 AI4Math Skill：拉取各仓库最新的 main 分支，同步已安装的 Skill 目录， +告诉我哪些有变化。 +``` + +```text +卸载 <包名> 这个 Skill，从本机 skills 目录移除它，保留克隆下来的仓库。 +``` + +### 几点提醒 + +- **Skill 只是说明书，不是运行环境。** 比如 SageMath Skill 需要本机真的装了 SageMath，Lean Skill 需要 Lean 4 和 mathlib，优化类 Skill 需要对应求解器。这些环境的安装通常也由 Skill 自己引导 agent 完成，但会先征求你同意。 +- **涉及下载、装环境、创建 conda 环境的操作，先看清楚再批准。** +- **需要账号登录的环节由你本人完成**，不要把 API key 交给 agent 写进仓库。 +- 装完可能需要重启 agent 才能发现新 Skill，agent 会告诉你。 + +--- + +## 二、应用程序 + +### VeryMath 智慧教材 + +个人部署的课程学习工作台。左边读教材 PDF，右边跟 Copilot 对话。导入自己的教材，连上一个 Coding Agent，就能针对任意章节提问、讲解、出题，并生成思维导图、知识图谱和 LaTeX 课件。所有数据留在你自己的电脑上（默认 `~/.course-copilot/`），用你自己的 Agent 账号。 + +**核心功能** + +- 教材解析：把选定页或章节整理为可读的数字教材，提取正文、公式、图片和目录 +- 讲解内容：对概念、定理、算法、例题做分层讲解，可主动配图 +- 知识点出题：生成带递进提示和参考答案的练习卡片 +- 思维导图 / 知识图谱：层级知识树与概念关系网络，节点可跳回教材原页 +- 生成课件：用 LaTeX Beamer 把章节编译成 16:9 的 PDF 课件 +- 自由问答：不带固定模板，按你的提问组织讲解和资料 +- 辅助资料：每本教材可挂讲义、习题解答、参考文献，支持全文搜索和扫描件 OCR + +**环境要求**：Node.js ≥ 22.13；生成课件需要本机有 XeLaTeX、Beamer、ctex 和中文字体；OCR 需要 Tesseract（中文还需 `chi_sim`/`chi_tra` 语言包）和 Poppler。Windows 建议整套装在 WSL2 里。 + +**安装**：同样是一句话交给 agent—— + +```text +拉取 https://github.com/VeryMath/VeryMath-textbook-copilot， +按仓库里的 skills/verymath-install/SKILL.md 完成 VeryMath 智慧教材的本机部署： +检查并安装必要依赖，配置内置课程 Skill，连接我现有的 Agent，启动工作台并给我访问地址。 +``` + +Agent 会装依赖、配好内置课程 Skill、连上 Agent 并启动服务，最后返回访问地址（正式运行默认 `http://127.0.0.1:4173`，只监听本机）。打开页面后点「导入教材」选 PDF，再到「工作区设置」连接 Agent 即可。 + +内置的课程 Skill 有 `textbook-parse`、`explain`、`quiz`、`mindmap`、`knowledge-graph`、`textbook-to-ppt`，加上负责部署的 `verymath-install`。 + +支持的 Agent：Codex、Claude Code、OpenCode、Cursor、Gemini CLI、Copilot CLI、Qwen Code、Kimi Code、Kiro CLI，选一种就行。 + +当前版本：v0.1.0（macOS 预览版）、v0.1.1（Windows 预览版）。 + +详细文档 → [VeryMath-textbook-copilot](https://github.com/VeryMath/VeryMath-textbook-copilot) ｜ [用户手册](https://github.com/VeryMath/VeryMath-textbook-copilot/blob/main/docs/user-guide.md) ｜ [架构参考](https://github.com/VeryMath/VeryMath-textbook-copilot/blob/main/docs/architecture.md) ｜ [Skill 开发](https://github.com/VeryMath/VeryMath-textbook-copilot/blob/main/docs/skill-development.md) + +--- + +### Co-Mathematician + +把数学研究对话变成有文件支撑的项目。核心思路是「Coding Agent + 仓库文件系统 + 门控 + 评审循环 = 研究工作台」：主线程扮演项目协调者，研究问题、目标、状态和消息都落在 `workspace/project/`，证明、计算、文献、评审各自成为独立工作流，最终论文只从通过评审的报告渲染出来。 + +Python 部分只负责初始化文件、追加消息、创建工作流、检查门控和渲染论文——它不跑 agent，agent 是你的 Coding Agent。 + +**核心机制** + +- 目标要标为 `approved` 才能开工作流，草稿目标不可执行 +- 工作流分 `proof`、`computation`、`literature`、`review` 四类 +- 每份报告要带来源溯源、显式的不确定性、失败尝试记录，以及 `reviews/` 下的独立评审 +- 最终产物 `workspace/final/working_paper.md` 是工作论文，不是聊天总结 +- 可以把 AI4Math 的领域 Skill 装进项目的 `.agents/skills/`，由协调者按需推荐和移交 + +**安装**：也是交给 agent—— + +```text +从 https://github.com/VeryMath/co-mathematician.git 安装 Co-Math， +把 ~/CoMathProjects 设为项目目录，创建一个名为 Muon Convergence 的项目。 +返回它的路径，先不要开始研究。 +``` + +然后用你的 Coding Agent 打开那个项目目录，说「Continue this Co-Math project.」即可开始。日常用 `co-math list / resume / next / archive / reopen` 管理项目。 + +详细文档 → [co-mathematician](https://github.com/VeryMath/co-mathematician) + +--- + +## 三、Skill 仓库一览 + +### AI4Math · SageMath — 精确符号计算 + +让 agent 用 SageMath 做可验证的精确计算：先检索内置的 SageMath 参考手册，再写成普通 Python 形式的 Sage 代码，本地执行，最后基于真实运行输出作答。覆盖代数、数论、组合、图论、多项式环、矩阵、微积分、有限域、椭圆曲线、Galois 群、编码理论、密码学、流形、模形式等领域。Windows 用户需要在 WSL 里装和运行 SageMath。 + + +| 包 | 用途 | +| ---------------- | ----------------------- | +| `sagemath-skill` | 参考检索 + 代码执行 + 结果复现的完整流程 | + + +详细文档 → [AI4Math-Sagemath-skill](https://github.com/VeryMath/AI4Math-Sagemath-skill) + +--- + +### AI4Math · Auto Research — 自动研究与问题发现 + +把模糊的研究意图变成结构化问题、有边界的研究运行和经过评审的证明计划。包最多的一个仓库，也包含几个外部研究系统（Rethlas、Archon、Danus）的安装与驱动助手。 + + +| 包 | 用途 | +| ----------------------------------- | -------------------------------------------------------------- | +| `discover-math-problems` | 把模糊背景转成排序过的问题、猜想格、证据账本和工单 | +| `proof-blueprint-review` | 构建证明蓝图，做验证者式评审，输出严格的证明状态报告 | +| `open-problem-research-pipeline` | 文献检索、方法分析、证明框架、报告的分阶段流程 | +| `graph-theory-researcher` | 读图论论文、经用户确认选题、有界研究、产出 LaTeX/PDF | +| `agent-laboratory-workflow` | 部署、校验并启动有边界的 Agent Laboratory 自动研究运行 | +| `rethlas-helper` | 安装 Rethlas 并从任意项目跑自然语言证明生成 | +| `archon-helper` | 安装 Archon 并跑形式化工作流 | +| `danus-helper` / `danus-helper-dsh` | 安装 Danus（OpenCode 版 / DeepSeek Harness 版），跑多智能体数学研究、监控、总结与论文写作 | + + +详细文档 → [AI4Math-Auto-Research](https://github.com/VeryMath/AI4Math-Auto-Research) + +--- + +### AI4Math · Paper Reading — 论文精读与转化 + +结构化地读数学论文，抽取定理依赖，并把论文转成可复用的 Skill。 + + +| 包 | 用途 | +| ---------------------------- | -------------------------------------- | +| `math-paper-reading` | 深读论文、抽取定理依赖、构建证明路径、管理本地文献笔记 | +| `paper-to-skill` | PDF 预处理、论文分诊、证明模式抽取、跨论文综合，产出新 Skill | +| `after-ocr` | 审校并修复公式密集的 OCR Markdown，带覆盖率、证据日志和多轮校对 | +| `graph-theory-paper-reading` | 图论论文深读，输出带证明树的结构化 LaTeX 报告 | + + +详细文档 → [AI4Math-Paper-Reading](https://github.com/VeryMath/AI4Math-Paper-Reading) + +--- + +### AI4Math · Writing — 论文写作与课件 + +从已验证的材料出发写论文，不替代证据核查本身。强调「有据可溯」：每个论断都要能追到来源。 + + +| 包 | 用途 | +| --------------- | --------------------------------------------------------- | +| `paper-writing` | 起草、修订、结构化和审计数学论文，含论断-证据核对、证明义务检查、记号一致性、公式可读性和 LaTeX 投稿前检查 | +| `math-beamer` | 构建有据可溯的 Beamer 讲稿，选模板、编译、检查排版、维护幻灯片到来源的对照表 | + + +详细文档 → [AI4Math-Writing](https://github.com/VeryMath/AI4Math-Writing) + +--- + +### AI4Math · Lean Agents — Lean 4 形式化 + +给 agent 一套检查 Lean 项目、搭 mathlib 工作区、形式化定理陈述、修复证明、补 `sorry` 并本地验证补丁的流程。最终补丁不允许引入 `sorry`、`admit`、新公理或悄悄改动定理陈述。 + + +| 包 | 用途 | +| -------------------- | --------------------------------------------- | +| `lean-setup` | 安装或检查 Lean 4、`elan`、`lake`，准备可复用的 mathlib 工作区 | +| `lean-formalization` | 形式化定理陈述、修复证明、补 `sorry`、评审 Lean 补丁、协调可选后端 | + + +`skills/lean-runtime/` 是共享支撑层（脚本、schema、提示词、测试），不单独调用，但安装时要保留在同级目录。Numina、Archon 等外部后端是可选适配器，需要你明确批准才会接入。 + +详细文档 → [AI4Math-Lean-Agents](https://github.com/VeryMath/AI4Math-Lean-Agents) + +--- + +### AI4Math · Optimization — 优化建模与求解 + +数学规划建模、求解器选型与环境配置、运行证据报告。 + + +| 包 | 用途 | +| ------------------------------- | ------------------------------------ | +| `optskills` | 103 张已发布的 OptSkills 问题原型卡，可独立用于建模和求解 | +| `mixed-integer-programming` | 含 0-1、整数、连续决策变量的 MILP/MIP 建模 | +| `second-order-cone-programming` | SOCP 建模与基于 cvxpy 的锥优化求解 | +| `osqp-solver` | 连续凸 QP 的 OSQP 建模、重复求解、状态门控与独立验证 | +| `cdopt-optimization` | CDOpt 与流形约束优化的建模、校验、runner 生成与证据报告 | +| `or-solver` | 指定求解器的依赖、安装、许可证、环境变量与故障排查 | + + +分工是：你或建模 Skill 选定求解器，`or-solver` 负责把那个求解器配好。 + +详细文档 → [AI4Math-Optimization](https://github.com/VeryMath/AI4Math-Optimization) + +--- + +### AI4Math · Computational Mathematics — 计算数学 + +把数学对象、方程、数据或论文片段变成经过评审的计算表示和可复现证据。 + + +| 包 | 用途 | +| ----------------------------------- | ------------------------------ | +| `invariant-computation` | 代数、拓扑、几何、TDA 与可认证数值不变量计算的路由与校验 | +| `least-squares` | 线性、多项式、非线性、正则化、带约束和贝叶斯最小二乘拟合 | +| `scientific-computing-reproduction` | 复现、诊断、调参、可视化和汇报科研计算代码，带人工审批检查点 | + + +详细文档 → [AI4Math-Computational-Mathematics](https://github.com/VeryMath/AI4Math-Computational-Mathematics) + +--- + +### AI4Math · Evolving — 演化式实验 + +目前聚焦有边界的 OpenEvolve 实验会话。 + + +| 包 | 用途 | +| -------------------------------- | ------------------------------------------- | +| `openevolve-experiment-workflow` | 检查或创建 OpenEvolve 项目、校验运行配置、跑短探针、汇总指标、指导迭代改进 | + + +详细文档 → [AI4Math-Evolving](https://github.com/VeryMath/AI4Math-Evolving) + +--- + +### AI4Math · MathTool — 独立小工具 + +放独立数学工具和轻量 agent 适配器，每个包有自己的工作流、依赖溯源和测试。 + + +| 包 | 用途 | 状态 | +| --------------- | ------------------------------------------- | --------------- | +| `math-glossary` | 构建、评审、备份、导入导出和维护双语数学术语表 | 已发布 v0.1.0（MIT） | +| `matlab-runner` | 把明确的 MATLAB 任务经 MATLAB MCP 服务器执行，带测试与数值验证证据 | 评审中，未发布 | + + +详细文档 → [AI4Math-MathTool](https://github.com/VeryMath/AI4Math-MathTool) + +--- + +## 四、官网 + +VeryMath 项目官网：组织介绍、仓库导航与动态。 + +- [VeryMath 官网](https://verymath.github.io/) +- [中文使用手册](/handbook/HANDBOOK.zh-CN.html) · [English Handbook](/handbook/HANDBOOK.html) +- [站点源码](https://github.com/VeryMath/verymath.github.io) diff --git a/index.md b/index.md index 9a18cf8..dce08e8 100644 --- a/index.md +++ b/index.md @@ -877,6 +877,36 @@ title: VeryMath

Updates动态

+
+
+
+ 2026.09 + Documentation文档 +
+
+

VeryMath Handbook

+

+ The bilingual handbook for VeryMath tool installation and usage is now online. + VeryMath 工具安装与使用方法 · 中英文手册正式上线。 +

+

+ Install Skills by talking to your coding agent, learn how standalone tools such as Textbook Copilot and Co-Mathematician work, and follow links to each repository’s detailed documentation. + 通过 Coding Agent 安装 Skill、了解智慧教材与 Co-Mathematician 等独立程序，并跳转各仓库详细文档。 +

+
+ What's inside包含内容 + + VeryMath tool installation and usage, a general Skill-install template you can send to your agent, application overviews, and AI4Math repository navigation. + VeryMath 工具安装与使用方法、可发给 Agent 的 Skill 安装通用模板、应用程序简介，以及 AI4Math 系列仓库导航。 + +
+
+
+
+ Handbook (中文)使用手册（中文） + Handbook (English)Handbook (English) +
+
diff --git a/tests/test_updates.py b/tests/test_updates.py index 5ea5fe2..f1b12af 100644 --- a/tests/test_updates.py +++ b/tests/test_updates.py @@ -5,6 +5,16 @@ EXPECTED_NEWEST_UPDATES = [ + { + "date": "2026.09", + "kicker": "VeryMath Handbook", + "category_en": "Documentation", + "category_zh": "文档", + "links": { + "/handbook/HANDBOOK.zh-CN.html", + "/handbook/HANDBOOK.html", + }, + }, { "date": "2026.09", "kicker": "EMNLP 2026 Findings · OptSkills × VeryMath", @@ -87,7 +97,7 @@ def test_updates_are_separate_and_newest_first(self): self.assertGreaterEqual(len(self.announcements), 6) for announcement, expected in zip( - self.announcements[:4], EXPECTED_NEWEST_UPDATES + self.announcements[:5], EXPECTED_NEWEST_UPDATES ): with self.subTest(kicker=expected["kicker"]): date = find_one( @@ -130,7 +140,7 @@ def test_updates_are_separate_and_newest_first(self): self.assertTrue(expected["links"].issubset(links)) def test_optskills_copy_names_the_paper_source_without_validation_claims(self): - optskills = self.announcements[0] + optskills = self.announcements[1] text = optskills.text() kicker = find_one( optskills, @@ -148,7 +158,7 @@ def test_optskills_copy_names_the_paper_source_without_validation_claims(self): self.assertNotIn("代表性验证", text) def test_optskills_footer_has_only_skill_and_paper_links(self): - optskills = self.announcements[0] + optskills = self.announcements[1] skill_link = find_one( optskills, lambda node: node.tag == "a" @@ -161,7 +171,7 @@ def test_optskills_footer_has_only_skill_and_paper_links(self): if node.tag == "a" and "href" in node.attrs } - self.assertEqual(links, EXPECTED_NEWEST_UPDATES[0]["links"]) + self.assertEqual(links, EXPECTED_NEWEST_UPDATES[1]["links"]) self.assertEqual( find_one(skill_link, lambda node: "lang-en" in node.classes).text(), "OptSkills", From 29c3c06e372a9707c8d89ca7666d1e00d0c65508 Mon Sep 17 00:00:00 2001 From: dyuan311 <98677639+dyuan311@users.noreply.github.com> Date: Fri, 18 Sep 2026 15:37:48 +0800 Subject: [PATCH 2/2] Announce the Textbook Copilot desktop app on the homepage. Add a bilingual Updates entry with Windows/macOS download links so visitors can find the installer from the site. Co-authored-by: Cursor --- index.md | 30 ++++++++++++++++++++++++++++++ tests/test_updates.py | 18 ++++++++++++++---- 2 files changed, 44 insertions(+), 4 deletions(-) diff --git a/index.md b/index.md index dce08e8..aa047a1 100644 --- a/index.md +++ b/index.md @@ -877,6 +877,36 @@ title: VeryMath

Updates动态

+
+
+
+ 2026.09 + Desktop App桌面应用 +
+
+

VeryMath Textbook Copilot

+

+ The VeryMath Textbook Copilot desktop app is now available. + VeryMath 智慧教材桌面 App 已上线。 +

+

+ A desktop app is now available on Windows and macOS. Download it, install it, and start using it right away. + 现已在 Windows 与 macOS 上推出桌面 App，下载安装，轻松使用。 +

+
+ What you can do能做什么 + + Parse textbooks, explain chapters, generate quizzes, mind maps, knowledge graphs, and LaTeX slides. One API key unlocks every feature. + 教材解析、章节讲解、出题、思维导图、知识图谱和 LaTeX 课件。一个 API Key 即可使用全部功能。 + +
+
+
+
+ Download下载安装包 + Repository仓库 +
+
diff --git a/tests/test_updates.py b/tests/test_updates.py index f1b12af..94b450b 100644 --- a/tests/test_updates.py +++ b/tests/test_updates.py @@ -5,6 +5,16 @@ EXPECTED_NEWEST_UPDATES = [ + { + "date": "2026.09", + "kicker": "VeryMath Textbook Copilot", + "category_en": "Desktop App", + "category_zh": "桌面应用", + "links": { + "https://github.com/VeryMath/VeryMath-textbook-copilot/releases", + "https://github.com/VeryMath/VeryMath-textbook-copilot", + }, + }, { "date": "2026.09", "kicker": "VeryMath Handbook", @@ -97,7 +107,7 @@ def test_updates_are_separate_and_newest_first(self): self.assertGreaterEqual(len(self.announcements), 6) for announcement, expected in zip( - self.announcements[:5], EXPECTED_NEWEST_UPDATES + self.announcements[:6], EXPECTED_NEWEST_UPDATES ): with self.subTest(kicker=expected["kicker"]): date = find_one( @@ -140,7 +150,7 @@ def test_updates_are_separate_and_newest_first(self): self.assertTrue(expected["links"].issubset(links)) def test_optskills_copy_names_the_paper_source_without_validation_claims(self): - optskills = self.announcements[1] + optskills = self.announcements[2] text = optskills.text() kicker = find_one( optskills, @@ -158,7 +168,7 @@ def test_optskills_copy_names_the_paper_source_without_validation_claims(self): self.assertNotIn("代表性验证", text) def test_optskills_footer_has_only_skill_and_paper_links(self): - optskills = self.announcements[1] + optskills = self.announcements[2] skill_link = find_one( optskills, lambda node: node.tag == "a" @@ -171,7 +181,7 @@ def test_optskills_footer_has_only_skill_and_paper_links(self): if node.tag == "a" and "href" in node.attrs } - self.assertEqual(links, EXPECTED_NEWEST_UPDATES[1]["links"]) + self.assertEqual(links, EXPECTED_NEWEST_UPDATES[2]["links"]) self.assertEqual( find_one(skill_link, lambda node: "lang-en" in node.classes).text(), "OptSkills",