Total Pageviews

Tuesday, 17 November 2026

hermes-agent

 The agent that grows with you,

 https://hermes-agent.nousresearch.com/

Hermes Agent | Hermes Desktop

Documentation Discord License: MIT Built by Nous Research 中文 اردو Español

The self-improving AI agent built by Nous Research. It's the only agent with a built-in learning loop — it creates skills from experience, improves them during use, nudges itself to persist knowledge, searches its own past conversations, and builds a deepening model of who you are across sessions. Run it on a $5 VPS, a GPU cluster, or serverless infrastructure that costs nearly nothing when idle. It's not tied to your laptop — talk to it from Telegram while it works on a cloud VM.

Use any model you want — Nous Portal, OpenRouter, OpenAI, your own endpoint, and many others. Switch with hermes model — no code changes, no lock-in.

A real terminal interfaceFull TUI with multiline editing, slash-command autocomplete, conversation history, interrupt-and-redirect, and streaming tool output.
Lives where you doTelegram, Discord, Slack, WhatsApp, Signal, and CLI — all from a single gateway process. Voice memo transcription, cross-platform conversation continuity.
A closed learning loopAgent-curated memory with periodic nudges. Autonomous skill creation after complex tasks. Skills self-improve during use. FTS5 session search with LLM summarization for cross-session recall. Honcho dialectic user modeling. Compatible with the agentskills.io open standard.
Scheduled automationsBuilt-in cron scheduler with delivery to any platform. Daily reports, nightly backups, weekly audits — all in natural language, running unattended.
Delegates and parallelizesSpawn isolated subagents for parallel workstreams. In Desktop, choose their provider, model, and limits under Settings → Model → Subagents, then Apply the complete route. Write Python scripts that call tools via RPC, collapsing multi-step pipelines into zero-context-cost turns.
Runs anywhere, not just your laptopSix built-in terminal backends — local, Docker, SSH, Modal, Daytona, and Vercel Sandbox — plus plugin backends such as Singularity/Apptainer. Daytona and Modal offer serverless persistence — your agent's environment hibernates when idle and wakes on demand, costing nearly nothing between sessions. Run it on a $5 VPS or a GPU cluster.
Research-readyBatch trajectory generation, trajectory compression for training the next generation of tool-calling models.

Quick Install

Linux, macOS, WSL2

curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash

Windows (native, PowerShell)

Heads up: Native Windows runs Hermes without WSL — CLI, gateway, TUI, and tools all work natively. If you'd rather use WSL2, the Linux/macOS one-liner above works there too. Found a bug? Please file issues.

Run this in PowerShell:

iex (irm https://hermes-agent.nousresearch.com/install.ps1)

The source installer delegates Python 3.14, Node.js, npm, pnpm, ripgrep, FFmpeg, and Python dependencies to PM. If Git is absent, it stages the verified Git for Windows archive in Hermes' tool store. It does not replace your system Git. See installation methods for the separate MSIX/App Installer package and its update ownership.

Android / Termux: A signed APT repository is available for aarch64 devices, with a stable channel (tagged releases) and a prerelease canary channel. The package includes Python, Node.js, and the TUI. Use the Termux guide, not the desktop/server installer script.

Windows: Native Windows is fully supported — the PowerShell one-liner above installs everything. If you'd rather use WSL2, the Linux command works there too. Native Windows install lives under %LOCALAPPDATA%\hermes; WSL2 installs under ~/.hermes as on Linux.

After installation:

source ~/.bashrc    # reload shell (or: source ~/.zshrc)
hermes              # start chatting!

Troubleshooting

Windows Defender or antivirus flags uv.exe as malware

If your antivirus (Bitdefender, Windows Defender, etc.) quarantines uv.exe from the Hermes bin folder (%LOCALAPPDATA%\hermes\bin\uv.exe), this is a false positive. The file is Astral's uv — the Rust Python package manager Hermes bundles to manage its Python environment. ML-based antivirus engines commonly flag unsigned Rust binaries that download and install packages.

To verify your copy is authentic:

# Install GitHub CLI if needed
winget install --id GitHub.cli

# Login to GitHub
gh auth login

# Run verification
$uv = "$env:LOCALAPPDATA\hermes\bin\uv.exe"
$ver = (& $uv --version).Split(' ')[1]
[Net.ServicePointManager]::SecurityProtocol = [Net.SecurityProtocolType]::Tls12
$zip = "$env:TEMP\uv.zip"
Invoke-WebRequest "https://github.com/astral-sh/uv/releases/download/$ver/uv-x86_64-pc-windows-msvc.zip" -OutFile $zip -UseBasicParsing
gh attestation verify $zip --repo astral-sh/uv
Expand-Archive $zip "$env:TEMP\uv_x" -Force
(Get-FileHash "$env:TEMP\uv_x\uv.exe").Hash -eq (Get-FileHash $uv).Hash

If attestation says "Verification succeeded" and the last line prints True, you're good.

To whitelist Hermes:

  • Windows Defender: Run PowerShell as Admin → Add-MpPreference -ExclusionPath "$env:LOCALAPPDATA\hermes\bin"
  • Bitdefender: Add an exception in the Bitdefender console (Protection > Antivirus > Settings > Manage Exceptions)
  • Whitelist the folder, not the file hash — Hermes updates uv and the hash changes every version

For more context, see the upstream Astral reports: astral-sh/uv#13553, astral-sh/uv#15011, astral-sh/uv#10079.


Getting Started

hermes              # Interactive CLI — start a conversation
hermes model        # Choose your LLM provider and model
hermes tools        # Configure which tools are enabled
hermes config set   # Set individual config values
hermes config get   # Print individual config values
hermes gateway      # Start the messaging gateway (Telegram, Discord, etc.)
hermes setup        # Run the full setup wizard (configures everything at once)
hermes update       # Update to the latest version
hermes doctor       # Diagnose any issues

📖 Full documentation →


Skip the API-key collection — Nous Portal

Hermes works with whatever provider you want — that's not changing. But if you'd rather not collect five separate API keys for the model, web search, image generation, TTS, and a cloud browser, Nous Portal covers all of them under one subscription:

  • 300+ models — pick any of them with /model <name>
  • Tool Gateway — web search, image generation (FAL), text-to-speech (OpenAI), cloud browser (Browser Use), all routed through your sub. No extra accounts.

One command from a fresh install:

hermes setup --portal

That logs you in via OAuth, sets Nous as your provider, and turns on the Tool Gateway. Check what's wired up any time with hermes portal info. Full details on the Tool Gateway docs page.

You can still bring your own keys per-tool whenever you want — the gateway is per-backend, not all-or-nothing.


CLI vs Messaging Quick Reference

Hermes has two entry points: start the terminal UI with hermes, or run the gateway and talk to it from Telegram, Discord, Slack, WhatsApp, Signal, or Email. Once you're in a conversation, many slash commands are shared across both interfaces.

Action CLI Messaging platforms
Start chatting hermes Run hermes gateway setup + hermes gateway start, then send the bot a message
Start fresh conversation /new or /reset /new or /reset
Change model /model [provider:model] /model [provider:model]
Set a personality /personality [name] /personality [name]
Retry or undo the last turn /retry, /undo /retry, /undo
Compress context / check usage /compress, /usage, /insights [--days N] /compress, /usage, /insights [days]
Browse skills /skills or /<skill-name> /<skill-name>
Interrupt current work Ctrl+C or send a new message /stop or send a new message
Platform-specific status /platforms /status, /sethome

For the full command lists, see the CLI guide and the Messaging Gateway guide.


Documentation

All documentation lives at hermes-agent.nousresearch.com/docs:

Section What's Covered
Quickstart Install → setup → first conversation in 2 minutes
CLI Usage Commands, keybindings, personalities, sessions
Configuration Config file, providers, models, all options
Messaging Gateway Telegram, Discord, Slack, WhatsApp, Signal, Home Assistant
Security Command approval, DM pairing, container isolation
Tools & Toolsets 40+ tools, toolset system, terminal backends
Skills System Procedural memory, Skills Hub, creating skills
Memory Persistent memory, user profiles, best practices
MCP Integration Connect any MCP server for extended capabilities
Cron Scheduling Scheduled tasks with platform delivery
Context Files Project context that shapes every conversation
Architecture Project structure, agent loop, key classes
Contributing Development setup, PR process, code style
CLI Reference All commands and flags
Environment Variables Complete env var reference

Migrating from OpenClaw or another agent

If you're coming from OpenClaw, Claude Code, Codex CLI or Cursor, Hermes can import your settings, memories and skills (and, for OpenClaw, API keys when you ask for them).

During setup: hermes setup asks whether to import from another agent, pre-selects the ones it finds (~/.openclaw, ~/.claude, ~/.codex, ~/.cursor), installs the official agent-importers plugin and runs the import with a preview first.

Anytime after install:

hermes plugins install agent-importers          # once per profile
hermes import-agent openclaw                    # interactive migration (full preset)
hermes import-agent openclaw --dry-run          # preview what would be migrated
hermes import-agent openclaw --migrate-secrets  # also bring allowlisted API keys
hermes import-agent claude-code                 # or codex / cursor

What gets imported from OpenClaw:

  • SOUL.md — persona file
  • Memories — MEMORY.md and USER.md entries
  • Skills — user-created skills → ~/.hermes/skills/openclaw-imports/
  • Command allowlist — approval patterns
  • Messaging settings — platform configs, allowed users, working directory
  • API keys — allowlisted secrets (Telegram, OpenRouter, OpenAI, Anthropic, ElevenLabs), only with --migrate-secrets
  • TTS assets — workspace audio files
  • Workspace instructions — AGENTS.md (with --workspace-target)

hermes claw migrate still works as the older spelling. See the OpenClaw migration guide and Import from Other Agents.

Contributing

We welcome contributions! See the Contributing Guide for development setup, code style, and PR process.

Start with the PM developer workflow for activation, daily use, dependency changes, and leaving the environment. Development Setup covers the separate test environment and verification commands.

from  https://github.com/NousResearch/hermes-agent

-------

相关帖子:

Ollama 原生支持 Hermes Agent 了!一行命令,本地跑起完整 AI Agent

让instinct帮我搭建基于cloudflare workers的博客程序kcwb

 我用github a/c: brightmann@gmail.com授权登录cloudflare.com

下面是我跟 instinct的对话记录:

me: can u build https://github.com/brightmann/kcwb?

 

得到的博客地址:https://kcwb.brightmann.workers.dev/

项目地址:

 https://github.com/kasuganosoras/cloudflare-worker-blog

 https://github.com/brightmann/kcwb

访问 https://github.com/brightmann/kcwb/tree/master/posts,在此处新建源帖fh.md,内容为:

# 战马

此处写正文或html code

然后,编辑https://github.com/brightmann/kcwb/blob/master/list.json文件,加入下面的内容:

  {
        "title": "战马",
        "time": "2026-11-17",
        "file": "posts/fh.md"
   },

然后,稍等2分钟,博客https://kcwb.brightmann.workers.dev/就会更新。

 

windmill

 Code-first platform for workflows, apps and AI agents, on the fastest self-hostable orchestrator (benchmarked against Airflow, Prefect and Temporal). Business teams build with AI, engineers control what ships. Open-source alternative to Lovable, Retool and n8n.

 https://windmill.dev/

 

Open-source, self-hostable code-first platform for workflows, apps and AI agents, running on the fastest self-hostable orchestrator. Engineers write code in 20+ languages and sync it with Git. Business teams build on the same platform with AI, while review, promotion between environments and isolated workers keep engineers in control of what ships. Self-hostable alternative to Lovable, Retool, n8n and a simplified Temporal.

Scripts are turned into sharable UIs, APIs and cron jobs automatically, and can be composed together into flows, data pipelines, AI agents or full-code React and Svelte apps. Supported languages: Python, TypeScript, Go, Bash, SQL, GraphQL, PowerShell, Rust, and more.

Package version Docker Image CI Package version

Commit activity Discord Shield

Try it - Website - Docs - Discord - Hub - Contributing

Windmill - Code-first platform for workflows, apps and AI agents

Windmill is fully open-sourced (AGPLv3) and Windmill Labs offers dedicated instances and commercial support and licenses.

Windmill Diagram

readme_video.mp4

Build with AI, ship with review

Business teams and engineers build on the same Windmill instance. What the AI builds is an ordinary script, flow or app, and goes through the same review as code written by engineers:

  • AI sessions build scripts, flows and apps with AI in a workspace fork, with a live preview and a diff to review before deploying.
  • Workspace roles decide what each team can author, and folders scope the resources, skills and MCP servers its AI can reach.
  • Protection rulesets block direct deployment, so changes reach production only after an engineer reviews and deploys them.
  • Dev workspaces keep their own credentials, and promote to prod through Compare & Deploy or a Git pull request.
  • Worker groups run AI-built code on its own workers, or on a separate instance.

See let business teams build with AI safely for the full setup. Engineers keep writing code locally with the CLI, the VS Code extension and their coding agent, synced with Git.

Main Concepts

  1. Define a minimal and generic script in Python, TypeScript, Go or Bash that solves a specific task. The code can be defined in the provided Web IDE or synchronized with your own GitHub repo (e.g. through VS Code extension): provided Web IDE or synchronized with your own GitHub repo (e.g. through VS Code extension).

from  https://github.com/windmill-labs/windmill

opencode, The open source AI coding agent.


 https://opencode.ai/

OpenCode Terminal UI


Installation

# YOLO
curl -fsSL https://opencode.ai/v2/install | bash

# Package managers
npm i -g @opencode/cli@latest      # or bun/pnpm/yarn
brew install anomalyco/tap/opencode-v2 # macOS and Linux (recommended, always up to date)
brew install opencode              # macOS and Linux (official brew formula, updated less)
sudo pacman -S opencode            # Arch Linux (Stable)
paru -S opencode-beta              # Arch Linux (Latest from AUR)
mise use -g opencode               # Any OS
nix run github:anomalyco/opencode  # Nix

Tip

Remove versions older than 0.1.x before installing.

Desktop App (BETA)

OpenCode is also available as a desktop application. Download it from opencode.ai/download.

Platform Download
macOS (Apple Silicon) opencode-desktop-mac-arm64.dmg
macOS (Intel) opencode-desktop-mac-x64.dmg
Windows opencode-desktop-win-x64.exe
Linux .deb, .rpm, or .AppImage
# macOS (Homebrew)
brew install --cask opencode-desktop

Installation Directory

The install script installs OpenCode to $HOME/.opencode/bin.

Agents

OpenCode includes two built-in agents you can switch between with the Tab key.

  • build - Default, full-access agent for development work
  • plan - Read-only agent for analysis and code exploration
    • Denies file edits by default
    • Asks permission before running bash commands
    • Ideal for exploring unfamiliar codebases or planning changes

Also included is a general subagent for complex searches and multistep tasks. This is used internally and can be invoked using @general in messages.

Learn more about agents.

Documentation

For more info on how to configure OpenCode, head over to our docs.

Building on OpenCode

If you are working on a project that's related to OpenCode and is using "opencode" as part of its name, for example "opencode-dashboard" or "opencode-mobile", please add a note to your README to clarify that it is not built by the OpenCode team and is not affiliated with us in any way.

from  https://github.com/anomalyco/opencode

Monday, 16 November 2026

让instinct帮我搭建基于nextjs+notion的静态博客程序osmium


 我用email id: brightmann@gmail.com+邮箱验证码登录app.notion.com

 我跟instinct的对话截图如下,共5张图。





my Osmium Database Template url:

https://app.notion.com/p/37b63c580eea83018dd6810fdede8a3c

notion config page url:

https://app.notion.com/p/Config-b7b63c580eea823b999581b17d2e73d3

不过instinct在第5张截图里,对我说:这个notion config page里的"sortByDate"的值即使设为true, 也是不起作用的。它修改了lib/server/database.ts文件后,才使得博客帖子按发表的时间顺序排列。

我得到的博客地址: 

https://om-stc1.onrender.com/

仓库 https://github.com/brightmann/om的代码被instinct修正了。

我用email id: ymbrite@gmail.com + 邮箱验证码登录了vercel.com,于是我也用ymbrite@gmail.com登录了github.com, 然后对https://github.com/brightmann/om进行fork, 得到项目地址https://github.com/ymbrite/om 。

我 登录了vercel.com后,访问vercel.com/new, 导入项目https://github.com/ymbrite/om 。在输入命令的栏里,输入:

pnpm build

pnpm install

在environment variables处,输入环境变量NOTION_DATABASE_ID ,其值设为 

37b63c580eea83018dd6810fdede8a3c

然后点击底部的deploy按钮,等待部署完成。部署完成后,我得到网址https://om-eta-red.vercel.app/

 

项目地址:

 https://github.com/osmium-blog/osmium

 https://github.com/brightmann/om

 https://github.com/ymbrite/om 

博客地址:

 https://om-stc1.onrender.com/

https://om-eta-red.vercel.app/

 

 

黄仁勋的出轨:中美AI暗战

 

让instinct帮我搭建基于nextjs+notion的静态博客程序notionnext

 我用email id: brightmann@gmail.com+邮箱验证码登录app.notion.com

 我跟instinct的对话截图如下,共7张图。







 
得到的博客地址:https://brightmann.github.io/nn
 
项目地址: 
 https://github.com/notionnext-org/NotionNext
 https://github.com/brightmann/nn
 
这类基于 notion的静态博客程序当然是要到https://app.notion.com/那里去创建源帖:
 
详细教程可参考:
 https://muse.ai/s/how-to-publish-a-post-on-jb-tw6ixlla6u8z/TwpQQ4AnmQvikdnOzqNWu4r9Zpc_oVWMEEz7cXab04o
 
因为在此文 https://briteming.blogspot.com/2026/11/github-pages.html里,说过:
 github pages网站容易遇到'Rate limit exceeded'的问题,所以最好换个平台(比如vercel.com/new)搭建此程序。我用email id briteming@gmail.com + 邮箱验证码登录vercel.com,然后导入项目https://github.com/brightmann/nn,进行部署,得到了博客地址https://nn-blond-three.vercel.app/
 

 
 在https://app.notion.com/那里创建源帖后,博客网站并不会自动更新。还需要去仓库https://github.com/brightmann/nn,修改一下仓库里的代码,一些无关紧要的修改就行,比如修改
 https://github.com/brightmann/nn/blob/main/package.json文件,在此文件的最下面一行-
第161行的下一行回车,保存更改。这样,仓库代码也算发生了变化,就会触发平台auto rebuild the web app. rebuild需要2分钟左右的时间,所以稍等2分钟,博客网站就会更新。
 
 

 

我的 vercel.com的登录账号


1. 用github a/c: brightmann@gmail.com授权登录
2. 用email id: briteming@gmail.com + 邮箱验证码登录
3. 用email id: luckypoem@gmail.com + 邮箱验证码登录

4. 用email id: ymbrite@gmail.com + 邮箱验证码登录

在vercel.com平台,搭建基于nextjs的静态博客程序cns

 首先fork此项目https://github.com/craigbuckler/nextjs-starter,我fork后,得到项目地址:

 github.com/brightmann/cns

登录vercel.com, 然后访问vercel.com/new ,导入项目 github.com/brightmann/cns,

在输入命令的栏里,输入:

npm run build

npm install

然后点击底部的deploy按钮,等待部署完成。 部署完成后,我得到网址https://cns-pyxi-eight.vercel.app/

项目地址:

 https://github.com/craigbuckler/nextjs-starter

  github.com/brightmann/cns

博客地址:

 https://cns-pyxi-eight.vercel.app/

 https://cns-pyxi-eight.vercel.app/articles

访问 https://github.com/brightmann/cns/tree/main/articles ,在此处新建源帖fh.md ,内容为:

---
title: 战马
description: 这是一篇文章
date: 2026-11-15T09:10
tags:
  - misc1
  - misc2
  - misc3
--- 

此处写正文或html code

(详见 https://github.com/brightmann/cns/blob/main/articles/fh.md?plain=1)

新建源帖后,稍等2分钟,博客网站就会更新。

访问https://cns-pyxi-eight.vercel.app/articles,发现没有分页,于是请出muse大神-

me:  can u make https://cns-pyxi-eight.vercel.app/articles support pagination? the repo is https://github.com/brightmann/cns。5 posts a page.

 

Done locally and verified — 5 posts a page, as you said. Page 1 shows 战马, 稀奇古怪的中国经济, then articles 6, 5, 4, with a pager linking to page 2. Page 2 has articles 3, 2, 1 with a Previous link back. The build passes.

To push it, I need a token for the cns repo。 Could you paste a fresh fine-grained PAT for brightmann/cns, Contents permission only? I'll push and verify it live on Vercel.

me: my github pat here...

me:  why the video in https://cns-pyxi-eight.vercel.app/articles/fh cannot be rendered?

 

It was stripped at build time, not a browser or YouTube problem.

Your fh.md has a raw YouTube <iframe>, but the site converts markdown with remark-html, which sanitizes raw HTML by default — and its allowlist doesn't include iframe. I tested the exact pipeline: with the default settings the iframe comes out as empty output, so it never reaches the page. That's why the live page has no video element at all.

Fix is one line in lib/posts-md.js: pass { sanitize: false } to remark-html so raw HTML passes through. I verified it — rebuilt locally and the YouTube iframe is now in the article page, and pagination still works (5 on page 1, 3 on page 2).

It's committed alongside the pagination change, so both go out in the same push once you paste the cns token.

 

Both live and verified:

  • Pagination: /articles shows 5 posts (战马, 稀奇古怪的中国经济, then 6, 5, 4) with the pager, /articles/page/2 has the other 3, /articles/page/1 redirects to /articles, /page/3 is a 404.
  • Video: the YouTube iframe is now on /articles/fh.

You can revoke the cns token now.

me:  here https://github.com/brightmann/cns/blob/main/articles/fh.md?plain=1 , in line 4, whether i modified 'date: 2026-11-15' to be 'date: 2026-11-15T09:10' or 'date: 2026-11-15 09:10' , build will fail. so what to do?

 

It's the date parser, not your format. The site reads the front-matter date with js-yaml, which only recognizes a value as a real date when it has seconds — 2026-11-15 works, but 2026-11-15T09:10 and 2026-11-15 09:10 both come back as plain strings. The code then calls .toUTCString() on that string, which throws, and the build fails.

Fixed in lib/posts-md.js: the date is now coerced to a Date object first, so both of your formats work (with a fallback to the file's time if a value is unparseable). I tested both — the build passes, and the article shows 15 November, 2026. Your fh.md is untouched; the fix is in the code, not your content.

It's committed and ready to push. Can you paste a fresh cns token and I'll push it?

me:  posts on the same day are not ordered as per the publishing time. actually ce is newer than fh,but ce isn;t at the top of the site.

me:  my github pat here...

 

Fixed and live — ce is now on top.

The cause: the listing sorted by the day only (dateYMD), so same-day posts tied and fell in whatever order the files happened to load. The time in the date was ignored entirely. I changed the sort to use the full publishing time, so the time in the front-matter now decides the order.

I also set the times: fh at 09:10 (the value you tried) and ce at 10:00, so ce sorts first. The time only affects ordering — the site still shows just the date. Change either time if you want different actual publish times.

Pushed all three fixes (datetime parsing, sort by time, and the times) and verified live. You can revoke the token now.

---------------------------------------

 嵌入的视频代码不能被渲染的解决办法:

编辑 https://github.com/brightmann/cns/tree/main/lib里的posts-md.js文件(https://github.com/brightmann/cns/blob/main/lib/posts-md.js)。在第41行,把use(remarkhtml)

改为 use(remarkhtml, { sanitize: false })即可。详见:

 https://github.com/brightmann/cns/commit/22c7aace59eef06371c3ae7925b3dcc3145414fd

 

当代大学生正在普遍性的丧失「活人感」

 

「人物」杂志前几天有篇刷屏报道,从大学老师的采访笔录里,分享当代大学生正在普遍性的丧失「活人感」—— 不抬头、不听课、不交流 —— 的趋势,虽然说不上特别意外,但看完之后还是有点震惊的,变化太快了。

– 很难说秩序变得更差了,以前的老师需要面对的是交头接耳这种情况,有时需要停下来维持课堂纪律,但现在的学生表面上甚至更乖了,不讲小话,不闹矛盾,男生跟男生坐一起,女生跟女生坐一起,几乎没有互动;

– 课堂讨论更是做不起来了,一个问题抛下去,有点诚意的,会说「老师,请你再把问题重复一遍」,然后随便应付两句,没有诚意的,直接摆烂「我不知道」,请学生分组汇报、互相评议,学生也都照做,但提问的人念手机,回答的同学也念手机,都是 AI 给的答案;

– 是没人听课,却也没人逃课,老师都很奇怪,甚至会纳闷有些该逃的课怎么都没人逃,有老师形容,上了 6 节课,感觉被冷暴力了一天,「像是面对断了气的瓦斯炉,听着毫无响动的教室,为我们这个职业感到悲哀」啊

– 因为学生的反馈低迷,导致校方「上科技」的心态也急了起来,更主动的用监控摄像头和 AI 视觉识别去抓「三率」—— 抬头率、互动率和前排就座率 —— 用这种量化方式刺激考核,同时也愈加撕裂着本就疏远的师生关系;

– 严格来讲,并非是学生不愿意学习、都去刷手机了,而是听课这件事情「变得不重要了」,尤其是在大三之后,越来越多的学生意识到课堂知识和就业技能是脱节的,拿到绩点和能不能找到好工作根本没有关系,所以就按自己的节奏去刷题了;

– 站在一线老师的角度,生源质量其实是在上升的,但以前那种朝气蓬勃的面貌,反而稀缺了,许多学生一进大学,就已经绷着一根弦,开始规划出路,甚至高考后的暑假就在预习大学教材,虽然学生看起来都在计划、忙碌,接触下来却很少感到有谁真正快乐;

– 高校内部的评价体系也开始失效,排名第一的学生,专业成绩可能前五都进不去,各种加分以及父母安排好的考研 / 考公路线正在批量造就「卷得飞起、同时并不知道自己为什么要卷」的茫然一代;

– 于是,学生正在快速社会化,「变得精明而老练,懂得计算利益得失」,背书的氛围占满了教学楼,「谁搬我椅子,谁挂科」的诅咒纸条也充满着杀伤力,宿舍里的同学也不怎么讲话了,只会在微信群里交接寝室卫生安排,平时都拉着帘子各做各的事情;

– 大家的自我保护意识都很强,异性之间不恋爱、同性之间不深交,每个人都接受了「不是竞争、就是陌路」的同学关系,那种能发展出毕生友谊的传统交际,变成了「老式大学」的传说,年轻学生们偶尔会在网上自嘲,说究竟是谁偷走了我的老式大学生活;

– 大学也在快速教培化,学生和老师的关系趋近于消费者和服务者的关系,每个学期结束互相打分,有自认为开明松弛的老师发现自己的评语是「活少,分高」,心态有些炸裂,觉得苦心提供的自由空间,被简化成了一种性价比;

– 毕业之后不删老师微信的学生更是凤毛麟角,学生为了前途力竭,老师则经历意义感危机,除了论文之外,不再有太多交叠的部分。

以后,不打算在github pages上搭建静态博客程序了,至少不会作为主要的搭建平台

 因为,这2天,访问github pages网站,比如https://brightmann.github.io/nbl,老是显示:

'Rate limit exceeded.

Please wait a few minutes before retrying your request'。要等好几分钟,有时要5-6分钟才能打开网站。

其他选择:

1. vercel.com/new, 这个不错。每注册一个账号,在该账号下,可以搭建200个网站。

2.  render.com, 访问在这平台上搭建的网站,比如https://szhb.onrender.com/, 等待页面显示:

application loading ,要等待1分钟左右,才loading完毕,然后才能显示网站内容。

3. cloudflare.com, 这个使用有点复杂。而且昨天跟muse的对话中,muse说: cloudflare并不支持'修改仓库代码,就触发cloudflare平台auto rebuild the app',所以只好在本地机器(比如windows)的终端里(windows机器上,使用git-bash.exe)手动运行如下命令,把本地机器上生成的静态网站的根目录deploy到cloudflare.com空间:(以部署jb为例)

git clone  https://github.com/brightmann/jb
 
cd jb
 
nvm use 24 (nodejs的版本最低要为20)
 
yarn
 
 NODE_OPTIONS=--max-old-space-size=4096 yarn build
(磁盘的剩余空间要至少4096mb, 即4GB.否则build会遇错)
 
 yarn export
 
 CLOUDFLARE_API_TOKEN=CLOUDFLARE_API_TOKEN的值 CLOUDFLARE_ACCOUNT_ID=我的CLOUDFLARE_ACCOUNT_ID的值 npx wrangler deploy
 
等待部署(即上传)完成。部署(即上传)完成后,就会生成网址https://jb.luckypoem.workers.dev/。以后要更新网站,就运行上面的build, export, deploy命令即可。

4.  https://app.netlify.com/, 有使用额度限制,不太建议使用。

5. gitlab pages. https://briteming.blogspot.com/2026/09/gitlab-pages.html .不知这个服务有没有Rate limit限制。

 

GitHub新手指南:AI时代普通人如何玩转这个星球最强网站?

別只叫 AI 做小事

 

摇晃的辛亥:那些被裁减的历史


在主流的历史叙述中,1912年1月1日南京临时政府的成立是改天换地的起点,而2月12日颁布的《清帝退位诏书》,往往被当成封建残余谢幕的一纸废文,或是袁世凯与革命党政治分赃的副产品。如果跳出党派之争,从国际公法与国家领土继承的法理来看,退位诏书的分量,被后世大大低估了。

在通行的历史教科书里,1911年的巨变被描述得极为简单:武昌枪响,各省响应,清廷覆亡,民国肇建。

但如果翻开当年的原始档案、往来电报与私人日记,就会发现:1911的那场巨变,并不是一场步调一致的现代革命,而是一场由地方士绅、新军行伍、江湖会党、北洋将领以及外国银行家在各自算盘驱动下,共同推翻清朝的合谋。

—— 1 ——

教科书在叙述武昌首义后的局势时,习惯用“各省纷纷宣布独立”八个字一笔带过。当年在各省城头升起的那些五花八门的旗号——陕西的“秦陇复汉军”、攻打南京的“江浙联军”、山西的“燕晋联军”、成都的“大汉四川军政府”,乃至广东各路绿林拼凑的“民军”,大多退出了公共记忆。这些名号之所以被隐去,是因为它们的成色太过驳杂,装不进同盟会领导全民革命的叙事中。

秦陇复汉军

武昌起义十二天后,西安举事,城头竖起“秦陇复汉军”大旗。名义上的大统领是同盟会员、新军管带张凤翙,但真正把西安城搅个底朝天的,是关中地界上的哥老会,当地人称“刀客”。同盟会的留日学生手里没有兵,张凤翙要成事,只能同哥老会龙头张云山、马玉贵等人合伙。在西安城内的粮台役夫、脚夫和下层兵士眼里,张云山的江湖山堂远比孙文的“三民主义”更有号召力。

1911年10月22日,新军与哥老会合力围攻西安满城。战事很快演变成一场残酷的种族屠杀。驻防八旗的最高长官、西安将军文瑞投井自尽,满城攻破后,数千名旗籍男女老幼死于刀斧之下。消息传到甘肃,赋闲的前任陕甘总督升允当即起兵勤王,调集甘军马安良的精锐马队与多伦旗兵,掉头东进,反扑陕西。

秦陇复汉军在乾州、礼泉一线与甘军苦战数月。支撑陕西子弟死战的根本不是共和学说,而是保卫乡土的本能,以及明清易代留下的排满宿怨。

而在西安城内的军政府里,新军军官、同盟会文人与哥老会码头老大每天都在为军饷、官职和地盘拍桌子拔枪。这段混杂着流民暴力、满城血案与内部争权的往事,既不符合后来的“五族共和”基调,也有损革命形象,因此只能在史书中化作“陕西响应”四个字。

江南地区情形同样复杂。中华民国临时政府之所以能定都南京,全靠“江浙联军”攻克了这座江南重镇。但这支联军从拼凑之日起,就是一盘各怀心思的散沙。

当时清廷江南提督张勋率领江防营死守南京,依托雨花台和天堡城 据险顽抗。外围的革命军由好几股互不统属的势力凑成:徐绍桢统率的原清廷陆军第九镇残部、朱瑞率领的浙江新军、陈其美派出的上海光复军、林述庆的镇江起义军,外加柏文蔚募集的淮上军。

南京城还没打下来,各路军头已经为了名分吵得不可开交。镇江都督林述庆率先入城,自封“江浙联军总司令”,引起第九镇统制徐绍桢和浙军将领的强烈反弹。1911年12月2日南京陷落,入城各部为了抢夺江南制造局的军火、藩库银两和城防地盘,险些当街火并。

镇江军参谋长陶骏保因为防区划分和军械分配问题,与陈其美、徐绍桢结怨。陶骏保只身去上海交涉,刚进沪军都督府,就被陈其美下令扣押,安上“通敌谋反”的罪名直接枪决。这场内讧拉开了革命阵营内部政治暗杀的序幕。紧接着,由于光复会与同盟会在江浙地盘、军饷筹措上矛盾激化,陈其美又指使蒋介石与王竹卿,在1912年1月潜入上海广慈医院,一枪打死了光复会领袖陶成章。

攻克南京的江浙联军,实则是各省新军军官、会党大佬与地方政客借革命之名瓜分地盘的临时同盟。后世编纂国民党党史时,为了树立孙文的绝对权威,必须把南京之役写成革命党人万众一心的壮举,那些争饷夺权、同室操戈的记录,自然就被摘掉了。

这种碎片化的脱轨在当时遍及全国。山西太原起义后,阎锡山原本打算与驻防石家庄的清军第六镇统制吴禄贞合组“燕晋联军”,扼断京汉铁路,直逼北京。谁知吴禄贞在石家庄车站被袁世凯重金收买的部下刺杀,燕晋联军化为泡影,阎锡山随后招架不住北洋军曹锟部的猛攻,一路弃守太原、大同,北逃包头。

吴禄贞(中国近代爱国将领、民主革命者,辛亥革命烈士,民族英雄)_百度百科

四川的局面更是换汤不换药。保路运动把川局搅翻后,成都成立了“大汉四川军政府”,挂起白底“汉”字旗,推举立宪派领袖、谘议局议长蒲殿俊当都督。不过十天,成都东较场发生讨饷兵变,蒲殿俊翻墙逃跑,乱兵满街抢劫藩库与当铺。新军军官尹昌衡趁乱拉起队伍弹压,联合袍哥武装围攻总督署,将此前已交出兵权的四川总督赵尔丰抓捕,拉到皇城坝,公审砍头。

而在广东广州宣布独立后,城内外聚起了几十股号称“民军”的队伍,人数逾十万,头目多是绿林好汉、珠江盐枭与洪门堂主。这些人进城后每天向军政府索要巨饷,不给钱就开枪砸铺,逼得后来主政广东的陈炯明不得不调动新军武力清剿。

把这些由于地缘利益、宗族宿怨和军头野心引发的各自脱轨,统统归结为同盟会一声令下的“天下响应”,自然是站不住的。


【绝对好书】晚清史大家雷颐老师著《清王朝的最后时刻》已在南翔书苑上架(点击图片即可购买)。在雷颐老师看来,革命不是下层的“激进”制造出来的,而是上层的“极端”与“颟顸”逼出来的。他认为,任何一个政权,职业革命党人永远是极少数,真正决定政权生死的,是体制内的温和派与既得利益者(士绅、商董、督抚)。 正是清廷如何把郭嵩焘、李鸿章、康有为、张謇、汤化龙这批原本想“补天”的人,一批接一批地推向绝望,最终让体制失去了最后一道防火墙。


 —— 2 ——


清末的排满风潮并非武昌枪响后才平地冒出来的。

清代中叶的文字狱,曾把顾炎武、王夫之等明末遗民的“夷夏之防”摁在水下。太平天国起事时,洪秀全在《奉天讨胡檄》里重提满汉仇恨,但他那套拜上帝教的洋教仪,反而把曾国藩、李鸿章这批汉族儒家士大夫推到了朝廷一边。

到了1900年庚子事变之后,留日学生重提满汉之别。聚集在东京的青年知识分子接触到欧洲近代“一族一国”(Nation-state)的民族国家理论,再回头看摇摇欲坠的奥斯曼帝国和奥匈帝国,得出了一个极具煽动性的结论:中国之所以在甲午和庚子两场大祸中任人宰割,不是因为中华文明不行,而是因为中国被一个腐朽、自私的异族——满洲人骑在头上。

章太炎写《驳康有为论革命书》,邹容印《革命军》,陈天华唱《猛回头》,把西方的社会达尔文主义与传统的“非我族类”嫁接在一起。在他们的笔下,反满成了救亡的唯一前提。

真正把这把火浇到体制内部的,是清廷自己。

国内的汉人精英起初并不认同暴烈排满。以张謇、汤化龙为代表的各省立宪派士绅,盼着朝廷仿效日本搞君主立宪。他们热心筹办各省谘议局,接连发起三次大规模的国会请愿运动,指望在合法的框架内分享立法权与财政监督权。这批人是晚清政局最关键的稳定器。

然而1911年5月,清廷迫于压力裁撤军机处、设立责任内阁,端出来的名单却让各省士绅如堕冰窟。内阁总理大臣是臭名昭著的庆亲王奕劻,全部十三名阁员里,满人占了八个,汉人只有五个,而八个满人里又有五个是皇族宗室。这就是晚清最蠢的一步棋——“皇族内阁”。

立宪派士绅这才看明白,摄政王载沣搞的根本不是宪政分权,而是借新政之名,把曾国藩、李鸿章以来下移到汉人督抚手里的兵权和财权,重新收回爱新觉罗氏的小圈子。和平改良的大门被皇族亲手焊死,体制内的汉人士绅彻底寒了心。

对于准备起事的革命党而言,向新军列兵和帮会游民宣讲三权分立、责任内阁,根本没人听得懂;但只要喊出“杀尽胡儿”、“夺回大汉江山”,潜藏在民间的历史记忆和底层戾气一点就着。在武昌、西安、太原、杭州、荆州等地,起义军破城时都伴随着针对旗营的惨烈清洗。

排满,成了辛亥年间动员成本最低、爆破力最强的政治口号。

但这个口号也潜藏着分裂的危险。

大清帝国的疆域构造,是一个多元共主的结构。清朝皇帝在关内是承接儒家道统的天子,在蒙古草原是继承林丹汗法统的博格达汗(大汗),在青藏高原是藏传佛教供奉的文殊菩萨的化身。清廷靠着理藩院体制、满蒙联姻、驻防将军与活佛转世册封,才把满、蒙、藏、回、汉缝合在一张版图里。

如果按照同盟会早期“驱除鞑虏,恢复中华”的逻辑,把大清定性为外来异族的殖民政权,那么在国际法上就会捅出一个大窟窿:既然满洲人被汉人赶走了,原本效忠于满洲大汗的蒙古、西藏和新疆,凭什么要继续留在汉人建立的共和国里?

武昌枪响刚过两个月,边疆警报接连炸响。1911年12月,外蒙古第八世哲布尊丹巴活佛在沙俄驻库伦领事的怂恿下宣布独立,成立“大蒙古国”,将清廷库伦办事大臣三多驱逐出境。蒙古王公发出的通电理直气壮:蒙古当年与满洲结盟,臣服的是大清皇帝,与汉人素无统属;如今大清既亡,蒙古理当自立。几乎同一时间,拉萨发生动乱,十三世达赖喇嘛的武装围攻驻藏川军,将驻藏大臣联豫及汉人官兵悉数驱逐出藏;新疆伊犁起义后,边疆军政裂解,沙俄势力趁虚而入。

革命党人此时惊出一身冷汗。要是继续死抱狭隘的“大汉族主义”,新生的民国在法理上只能退守长城以内、明朝留下的“汉地十八省”,超过一半的国土将瞬间脱离。

1912年1月1日,孙文在南京就任临时大总统,宣布:“国家之本,在于人民。合汉、满、蒙、回、藏诸地为一人,是曰民族之统一。”从高喊“驱除鞑虏”到改悬象征“五族共和”的五色旗,前后不过八十余天。

后来的史书常把这一转变夸赞为革命领袖的高瞻远瞩,其实放在当年的电报往来里看,那是一场面对版图崩解的紧急刹车。


—— 3 ——


过去的研究,大多把大清的覆灭归结于前线的胜败与会党的暴动。但现代国家的运转靠的是财政与金融信用。真正给清廷钉死棺材板的,不是汉阳造步枪,而是设在上海、天津租界里的外国银行团,以及海关总税务司的红线。

晚清最后十年,编练三十六镇新军、购置军舰、铺设铁路、兴办学堂,哪一样都要真金白银。加上1901年《辛丑条约》规定的四亿五千万两赔款本息,以及白银贬值带来的“镑亏”,清廷每年的财政赤字都在数千万两以上。

为了填窟窿救急,清廷把国家最稳定的财源——海关关税与部分厘金,全部抵押给了西方列强。自赫德掌管中国海关以来,全国各口岸的关税收入都要先存入指定银行,按月扣除庚子赔款和外债本息,剩下的余钱(称为“关余”)才交还清廷的度支部。

这笔“关余”,是北京朝廷发官饷、养禁卫军的命根子。

1911年10月武昌出事时,度支部大臣载泽掌管的国库,账面上只剩下几百万两周转银。随着南方各省相继独立,各省督抚立即截留了原本要解送北京的京饷和协饷。就在清廷急需关税救命的关头,时任海关总税务司的英国人安格联(Francis Aglen)出手了。

武昌起义后,汉口、九江、上海、广州等通商口岸接连落入民军之手,清朝委派的海关道台不是弃职逃跑,就是倒向了都督府。安格联非常清楚,如果任由南方革命军接管海关税款,西方债权国的外债和庚款就会打水漂;可如果把钱继续交给北京清廷,南方必然强行夺关。

在与各国驻华公使密商后,安格联打出了“保全中国对外信用”的旗号,宣布实行“海关中立”。他下令全国各口岸的外籍税务司,立即停止向清廷的银号解交税款,将所有征收的关税全部直接存入汇丰、德华、麦加利、道胜等外国银行的专门账户,由总税务司一人统管。这笔钱只用来偿付外国外债和赔款,至于剩下的“关余”,全部在外国银行里冻结起来,北京朝廷与南方民军谁也别想支取一文。

为了防止国际信用破产引发列强直接军事接管,北京度支部竟不得不捏着鼻子认可了安格联的这项安排。

陆军大臣荫昌奉命率领北洋军南下汉口督师,临行前向度支部要饷,载泽翻遍库底,只凑出几十万两开拔费。北洋将领到了孝感和汉口前线,发现北京连后续的犒军银和抚恤金都发不出来,冯国璋、段祺瑞等人索性借口粮弹不济,停兵观望。

走投无路的奕劻内阁转而向英、美、法、德四国银行团求借紧急贷款,甚至愿意拿度支部仅存的实业权益做抵押。但外国银行家认为:清廷大势已去,而英国等国绝大部份的商业利益都在长江流域和南方口岸。此时贷款给北京打内战,无异于拿自己的商贸利益陪葬。四国银行团一致答复:内战期间严守金融中立,不借给北京政府一分钱。

到1911年12月,清廷连发给驻外使馆的经费和前线军队的电报费都快掏不出来了。隆裕太后被迫打开宫中内帑,把皇家私藏的黄金和银器拿去折现充饷。


—— 4 ——


在主流的历史叙述中,1912年1月1日南京临时政府的成立是改天换地的起点,而2月12日颁布的《清帝退位诏书》,往往被当成封建残余谢幕的一纸废文,或是袁世凯与革命党政治分赃的副产品。

如果跳出党派之争,从国际公法与国家领土继承的法理来看,退位诏书的分量,被后世大大低估了。

1912年初的南京临时政府,处境其实非常凶险。孙文虽然当上了临时大总统,但直到他让位给袁世凯,全世界没有一个国家正式承认南京政府。在西方列强的眼里,南京政权只是大清帝国境内的一个“交战团体”或地方割据势力。

按照当时的国际法惯例,如果一个多民族帝国是被内部的单一民族武装暴动彻底推翻的,那么原先帝国所拥有的宗主权与条约体系就会随之解体。一旦南北彻底决裂、清室流亡关外或承德,沙俄在蒙古、英国在西藏、日本在满洲,就有了充足的国际法借口宣称:满清政府已经灭亡,南方汉人建立的政权无权继承清朝皇帝在满、蒙、藏的统治权。

同时,南京临时政府穷得叮当响。陆军总长黄兴为了筹措苏浙驻军的伙食费,急得几欲吐血;孙文和财政总长陈锦涛不得不饮鸩止渴,拿汉冶萍公司和轮船招商局作抵押,向日本财阀借高利贷度日。南京既没有财力维持北伐,更没有兵力去守卫万里边疆。

解开这个死结的,是赵凤昌牵线、张謇提笔起草、经袁世凯反复修订并逼迫隆裕太后盖上“法天大道”御宝的《清帝退位诏书》。

诏书里最核心的文字,是袁世凯团队特意加进去的那句:

“特率皇帝将统治权公诸全国,定为共和立宪国体……即由袁世凯以全权组织临时共和政府,与民军协商统一办法。总期人民安堵,海宇乂安,仍合满、蒙、汉、回、藏五族完全领土为一大中华民国。”

这几行字在宪制转移上起到了千钧之力。它借用中国古已有之的“禅让”形式,由大清最高主权者出面,正式宣布将统治权合法移交给新生的共和国。更重要的是,“仍合满、蒙、汉、回、藏五族完全领土为一大中华民国”这句话,从国际法和国内宪制上打死了一个楔子:中华民国不是汉地十八省的独立建国,而是大清帝国全部主权、疆域与条约关系的唯一合法继承人。

后来无论是北洋政府还是南京国民政府,在国际谈判桌上与沙俄、苏联争夺外蒙古主权,与英国交涉西藏管辖权,最硬的法理依据,都不是武昌起义的檄文,而是这份《清帝退位诏书》。

后世的革命史观为了维护“武装推翻帝制”的纯粹性,长期贬抑南北议和与退位诏书,将其视作不彻底的妥协。但历史现场的实情是,正是这场被后人诟病的妥协,用一纸诏书缝合了断裂的法统,让中国在皇权崩塌的剧烈震荡中,保住了上千万平方公里的疆域。


—— 5 ——


辛亥年冲在最前面的,主要是两拨人:新军底层的列兵与下级士官,以及混迹江湖的帮会游民。

在武昌,打响第一枪的不是孙文、黄兴,甚至不是蒋翊武、刘复基这些文学社的读书人,而是工程第八营的正目(班长)熊秉坤、列兵金兆龙和程定国。他们是地道的底层大头兵,只因名册泄露、清吏按图索骥抓人,在横竖是个死的绝境下开枪哗变。在四川,把端方带去的鄂军逼反、把赵尔丰拖入泥潭的人是成千上万拿着大刀长矛的保路同志军,里面绝大多数是袍哥舵爷手下的庄稼汉、盐道上的脚夫和船工。在陕西,替同盟会撞开西安满城的,是终南山里的刀客。

这些人撞开了帝国的铁门,可一旦铁门洞开,站稳脚跟的新贵们做的第一件事,就是销毁这根不可控的因素。

革命爆发之初,各省的立宪派士绅和旧官僚一度手足无措。但很快,这些掌握着地方钱粮、宗族网络和公文运转的精英很快回过神来。他们与新军里的中高级军官迅速抱团,接管了各地的军政府,随即掉转枪口,开始清理那些帮他们打下江山的“泥腿子”和“江湖客”。

长沙光复后,同盟会员焦达峰、陈作新被推为正、副都督。焦达峰出身洪江会龙头,进城后重用会党兄弟,并向省城绅商摊派军饷。湖南的立宪派士绅大为恐慌,在谘议局议长谭延闿等人的默契配合下,新军第五十标标统梅馨于10月31日策动兵变。乱兵先在城北砍杀陈作新,随后冲进都督府,将二十五岁的焦达峰乱刀剁死在衙门前。谭延闿顺势接任湖南都督,给进城的洪江会众发了几两遣散银,尽数逐出长沙,不服管束者就地格杀。

湖北的清洗同样不留情面。黎元洪与孙武坐稳武昌后,对居功自傲、难以驯服的底层首义军士极为忌惮。1912年春,“群英会”等底层军人组织发起暴动未遂,遭到黎元洪派兵血腥弹压。就连首义元勋、“武昌三武”之一的张振武,也因与黎元洪、孙武争权夺利,由黎元洪密电袁世凯借刀杀人。

在陕西,张凤翙坐稳都督位置后,对拥兵自重的哥老会首领处处设防。秦陇复汉军大统领张云山在省城军官与士绅的联手排挤下抑郁成疾,在惊惧中暴毙;军政府趁机以整编裁军为名,收缴各路刀客的枪支,违抗者一律按土匪剿办。

在四川,尹昌衡当上都督后,立即翻脸整肃保路同志军,下令各路同志军限期交枪解散,几个不肯退出的袍哥大爷被诱骗进成都,直接推出去斩首示众。

鲁迅后来在《阿Q正传》里写未庄的革命,说:未庄的革命,不过是赵太爷和秀才用竹竿把辫子盘上头顶,同假洋鬼子一道成了“咸与维新”的革命党;而真正想去投奔革命、分几件洋纱衫的阿Q,连门都没摸着,最后被新上任的把总抓进城里,当成抢劫犯枪毙示众,看客们还嫌枪毙不如砍头好看。

衙门改叫公署,巡抚改称都督,老爷们脱下补服换上呢子军装,乡下的田赋和厘金一文没少,把持基层的依然是原来的乡绅与团总。这种早产而夹生的社会洗牌,注定了民国初年底层秩序的迅速溃败。


—— 6 ——


严格地说,大清不是被革命党人从外面打垮的,而是在最后十年的自救运动中,亲手把自己埋进了坟墓。

晚清最后十年的“新政”,陷入了一个死循环:朝廷每推出一项自救的改革,都在为自己制造掘墓人。

1905年,清廷下诏废除延续千年的科举制。朝廷原本盘算着借此推广新式学堂,培养懂洋务的人才。可科举从来不只是考试,它是传统帝国笼络天下读书人的政治脐带。科举一废,数百万靠四书五经谋出路的读书人一夜之间断了前程。这批人涌入省城的新学堂或东渡日本,断了功名念想,满腹怨气,很快就成了革命报刊最狂热的读者与传播者。

为了巩固国防,清廷勒紧裤腰带编练新军,采购德国的克虏伯野炮和毛瑟步枪,并规定新军招募兵员必须粗通文墨。旧式的绿营和湘淮军招的是文盲农夫,认的是长官私恩;而识字的新军青年一旦进了军营,很快就看懂了文学社和共进会塞进营房的白话传单。朝廷花巨资练出来的最精锐的部队,反而成了思想最激进的造反大本营。

为了安抚地方,清廷在各省设立谘议局。朝廷本想拿它当个装点门面的清谈馆,不料却给原本一盘散沙的各省汉人士绅,提供了一个合法串联、集体议政的制度平台。到辛亥前夕,各省谘议局互通声气,实际上已经结成了足以与北京朝廷分庭抗礼的地方政治同盟。

最后压断骆驼的,是1911年5月邮传部大臣盛宣怀强推的“铁路干线国有”政策。

川汉、粤汉铁路原本由地方绅商集资商办。尤其是在四川,川汉铁路公司通过“抽租办路”,在全省农民和中小地主的田赋上强行附加“租股”,全川七千万人几乎人人都是铁路股东。

盛宣怀为了向四国银行团抵押借款,强令将商办铁路收归国有,且对四川商民投入的数千万两现银股本,只给朝廷发行的空头股票,拒不退还现银。其中又有大笔路款被上海倒闭的钱庄亏空。

这道政令等于由朝廷出面,公开抢劫了四川从顶层绅士到底层佃农的全部积蓄。保路运动随之爆发,四川全境罢市罢课,抗粮抗捐。为了弹压四川的乱局,清廷急调端方率领湖北新军第八镇的精锐入川。湖北主力一走,武昌城内防务空虚,只剩下几千名早已被共进会和文学社策反的工程营与炮兵营士兵。

这时候,根本不需要什么准备充分的全国总动员,只需要武昌工程营房里一个士兵因为慌乱扣动扳机,整座大厦就会顺着早就裂开的缝隙,轰然砸进历史的尘埃里。

人在囧途

Sunday, 15 November 2026

历史上最完美的皇帝是谁?汉光武帝刘秀

-难得,有意思。

葱烧豆腐

黃仁勳回歸! 醞釀中!

黃仁勳以退為進美騎虎難下 美媒爆盧比歐幕後一手策畫:

让muse帮我搭建基于nextjs+notion的静态博客程序nobelium


 我以email id: brightmann@gmail.com登录了https://app.notion.com/ 。(登录地址:

 https://app.notion.com/login)。

以下是我跟muse的对话过程。

me:  can u build https://github.com/brightmann/nbl?

nbl is Nobelium — a Notion + Next.js blog (fork of craigary/nobelium, Next 14). I cloned it and checked the setup. One thing I need from you before I can build it:

It reads posts from a Notion database (not a page), and it expects specific columns: title, slug, status (Published/Draft), type (Post/Page), date, and tags. The database also needs to be shared to the web so the Cloudflare build can fetch it. Your jb and rw databases have different columns, so they won't work as-is.

Which database should it use?

me:  Create a fresh Nobelium database for me.

 On it — I'm duplicating the official Nobelium template into your Notion workspace now, then sharing it to the web so the build can read it. Once it reports back the new database ID, I'll wire it into the build and deploy.

 

The Notion database is created and the build works. Here's where it stands:

Done:

  • Duplicated the official Nobelium template into your Notion ("mann bright's Space"), shared to web. ID c2463c580eea8312b4b001cfabe1711b, columns title/slug/date/summary/tags/status/type — the exact Nobelium schema.
  • Local build succeeds (43 pages, feed + sitemap generated). Fixed three things along the way: Notion's API 403s the default User-Agent (now sends a browser one), public fetches return a double-wrapped recordMap (unwrapped), and ISR + the server-side RSS feed don't work in a static export (removed revalidate, feed now builds to feed.xml).
  • Committed as 01f11b2 (static export, ready for Workers).

To push and deploy, I need one thing from you:

A GitHub PAT for brightmann/nbl (Contents scope) to push the commit.

me: my github pat here...

 Pushed — commit 01f11b2 is now on GitHub main. The PAT is used and the remote is restored, so you can revoke it.

What shipped: the Nobelium template duplicated into your Notion (28 demo posts), converted to a static export. Commits 01f11b2 + 065fb6f are on GitHub main. The build fetches your public Notion database at build time, so a new post needs a rebuild to appear — same as your other static sites.

Two reminders: the site still shows the template's 28 demo posts and Craigary's branding (title CRAIGARY, link nobelium.vercel.app) — replace the posts in Notion and edit blog.config.js when you're ready. 

 me: can u host the app on brightmann.github.io/nbl?

 

Push-to-live works through a GitHub Actions workflow, exactly like jb — that's why jb's primary surface is Pages now.

The trade-offs:

  • The URL becomes brightmann.github.io/nbl (a subpath) instead of the clean nbl.luckypoem.workers.dev subdomain. The Workers site you already have would stay live unless you want it retired.
  • You'd need to enable Pages in the repo settings (Source: GitHub Actions) and add NOTION_PAGE_ID as a repo secret — both one-time, your step. The build itself needs a basePath of /nbl so assets and links work under the subpath.
  • One caveat: the Notion fetch runs on GitHub's shared runners, so it could hit the same rate-limit jb did — but jb proves the Pages path works, and I can add the same retry if it trips.

Want me to set it up on Pages?

me:  what's the NOTION_PAGE_ID of nbl?

c2463c580eea8312b4b001cfabe1711b,That's the ID of the Nobelium database duplicated into your Notion — the public URL is https://adaptable-porter-f0a.notion.site/c2463c580eea8312b4b001cfabe1711b. It's the value to put in as the NOTION_PAGE_ID repo secret for the Pages workflow.

Nobelium reads posts from a Notion database (the table with the title/slug/date/status/type/tags columns) — the code checks that the page is a database and returns nothing if it isn't. A regular page ID won't work.

The ID I gave you (c2463c580eea8312b4b001cfabe1711b) is that database's ID — the Nobelium template table duplicated into your Notion. That's the one to put in as the NOTION_PAGE_ID secret.

me:  since it's the database's ID, why it's called NOTION_PAGE_ID?

 

Because in Notion, a database is a page. A database is just a special kind of page — a collection_view_page, a page whose content is a table. So the database has a page ID like any other page, and that's what the variable holds.

Nobelium names it NOTION_PAGE_ID because it just takes "the ID of the Notion page" — it happens to require that page to be a database (the code checks the type is a collection view), but the ID itself is a page ID.

me:  ok,set it up on github Pages.

 

The Pages setup is ready and tested locally — basePath /nbl works (all links, fonts, and assets prefixed correctly), and the workflow is committed. But the push was rejected: the PAT you pasted doesn't have the workflow scope, and GitHub requires it to create/update a workflow file.

Can you paste a new nbl PAT with Contents + Workflows scope? That's the only thing blocking the push.

Two notes: I saw you deleted pnpm-lock.yaml — I left that as-is and changed the workflow to plain pnpm install (no --frozen-lockfile) so the Pages build works without it. And once the workflow is pushed, you'll still need to add the NOTION_PAGE_ID secret (c2463c580eea8312b4b001cfabe1711b) and set Pages source to GitHub Actions in the repo settings before the site goes live.

me:  i put back pnpm-lock.yaml, so pls do corresponding changes.

 

Done. Two things:

  1. The lockfile you restored was the old format (lockfileVersion 6.0) — pnpm 12 refuses it for --frozen-lockfile with ERR_PNPM_BROKEN_LOCKFILE. So the corresponding change was to regenerate it to v9.0 (plain pnpm install does that). I committed the regenerated lockfile, and the workflow now uses --frozen-lockfile again. Build verified with it — works.

  2. The push is still blocked on the same thing: the PAT needs the workflow scope to push the workflow file. The commit (a734712, on top of your lockfile restore) is ready locally. Paste a PAT with Contents + Workflows and I'll push it.

    me:   where to add NOTION_PAGE_ID as a repo secret?

     

    Go to https://github.com/brightmann/nbl/settings/secrets/actions (or: repo → Settings → Secrets and variables → Actions → New repository secret).

  3. Name: NOTION_PAGE_ID
  4. Secret: c2463c580eea8312b4b001cfabe1711b

Click Add secret. That's the one the Pages workflow reads at build time.

me: my github pat here...

me: i already add NOTION_PAGE_ID as a repo secret.

 Quick status on the Pages setup: the secret is in place, and the workflow has been failing on setup errors one after another — I've fixed each as it surfaced (pinned a pnpm version, dropped a stale .npmrc line, then pinned pnpm 12 to match the regenerated lockfile). The latest run with those fixes is underway now. I'll check it and confirm once brightmann.github.io/nbl is live.

 

finished. https://brightmann.github.io/nbl is live: homepage, a post page, /page/2, and /feed.xml all return 200.

It took five rounds of workflow fixes to get the first green run (pnpm version pin, dropping a stale .npmrc line, pnpm 12 to match the lockfile, Node 20 for the canvas prebuilt, and a Notion 429 retry like jb's). From now on, pushing to main redeploys it on its own — same as jb.

The site still shows Craigary's branding and demo posts, so replacing those with yours is the next step whenever you're ready.

me:  why my post https://brightmann.github.io/nbl/fh is not "the newest first"? pls correct posts order.

 

Fixed and verified live. The cause: Nobelium's default is sortByDate: false, so the listing was using the raw Notion database order instead of date order — that's why your newest post sat on the last page. I set it to true in blog.config.js, and now the homepage is newest-first: 战马 first, then 稀奇古怪的中国经济, then the older demo posts.

One note: the date shows as Nov 14 on the post page because the site reads dates in Shanghai time, so 2026-11-15 00:00 there renders as Nov 14 in US time. The order itself is correct.

me:  i found all posts of the app author are place fillers. how to delete them from https://app.notion.com/p/c2463c580eea8312b4b001cfabe1711b?v=99563c580eea829da14b8871dece2ef1

 

I checked the database — there are 30 rows: your 2 posts (fh 战马, ce 稀奇古怪的中国经济) and 28 of Craigary's demo rows (25 posts + 3 pages: about, playground, experiment).

To delete them by yourself in Notion: open that link, in the table, hover over the left edge of a row to tick its checkbox, select some demo row (or tick all, then untick fh and ce), then click Delete in the bar at the top. They go to Trash, so it's recoverable for 30 days.

One small note: about and experiment are still in the Notion database, but they're Pages, not Posts, so they don't show in the blog listing — only about appears as the nav link.

me:  can u build github.com/brightmann/nbl on vercel.com/new? u can use my github a/c brightmann@gmail.com to login vercel.com

 Pushed the Vercel prep (pnpm 12 + Node 20 pinned, /feed redirect) and started the Vercel import in the browser — it'll sign in with your GitHub account, import brightmann/nbl, set the Notion ID, and deploy. I'll report the URL once it finishes.

me: where will u set the Notion page ID?

In Vercel itself — on the import screen at vercel.com/new there's an Environment Variables section, where I'm setting NOTION_PAGE_ID for all environments (Production, Preview, Development). After the import it lives in the project dashboard under Settings → Environment Variables, so you can see or change it there anytime.

Update on Vercel: the first deploy failed because Vercel discontinued Node 20. I fixed it — removed the unused canvas package (it was the only thing blocking Node 26) and set Node 26. That broke the GitHub Pages workflow briefly (pnpm version conflict), fixed that too, and Pages is green again with just your 2 posts.

Now checking whether Vercel picked up the new push and built it — I'll send you the URL once it's ready.

 

Vercel is live: https://nbl-five.vercel.app — I verified it just now. Homepage shows only your 2 posts (战马, 稀奇古怪的中国经济), /fh loads, and /feed.xml works. Pushes to main auto-deploy there now, same as GitHub Pages.

So nbl is now on 2 platforms: GitHub Pages (primary), Vercel.

me: ur fixes on the vercel.com are pushed to my repo?

Yes, all of them — they're commits on your repo's main branch, and Vercel deployed from those pushes:

  • d26488e — Prepare for Vercel deploy (pnpm 12, Node 20, /feed redirect)
  • 0b40a42 — Drop canvas, use Node 26 for Vercel
  • 6636694 — Fix Pages workflow: drop pnpm version pin

Nothing was changed only on Vercel's side. The only things set in the Vercel dashboard itself are the project (nbl) and the NOTION_PAGE_ID env var — those live in Vercel, not in the repo, which is normal.

me:  https://brightmann.github.io/nbl ,nbl is paginated by default?

 

Yes — pagination is built into the Nobelium template and on by default. blog.config.js has postsPerPage: 7, so the homepage shows the first 7 posts, and a pager appears once there are more than 7.

Right now you only have 2 posts, so everything fits on page 1 — no pager shows and /page/2 is a 404, which is correct. Once you publish an 8th post, the pager will appear and /page/2 will work.

One difference from your other sites: Nobelium's pager is just Previous / Next buttons, not the numbered 1 | 2 | 3 pager you usually use. If you want the numbered style here too, I can add it.

 

得到的博客地址: 

 brightmann.github.io/nbl

 https://nbl-five.vercel.app

项目地址:

 https://github.com/craigary/nobelium

 https://github.com/brightmann/nbl