Scaffold static blog + auto-publish CI; default work to develop

- Add Node static-site generator (scripts/build.mjs) rendering posts/*.md
  to public/ via marked + gray-matter; npm run build/serve.
- Add sample post and .gitignore (node_modules/, public/).
- Replace placeholder CI with publish.yml: scheduled daily build +
  auto-deploy to GitHub Pages from main.
- Document develop-default branch policy in AGENTS.md and README;
  enable npm ci in .codex/setup.sh.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Author: 123

.codex/setup.sh

@@ -10,16 +10,10 @@ set -euo pipefail
 
 echo "[codex-setup] OpenBlog: preparing environment"
 
-# --- Node.js / static-site generator (uncomment once a package.json exists) ---
-# if [ -f package.json ]; then
-#   echo "[codex-setup] installing npm dependencies"
-#   npm ci || npm install
-# fi
+# --- Node.js / static-site generator ------------------------------------------
+if [ -f package.json ]; then
+  echo "[codex-setup] installing npm dependencies"
+  npm ci || npm install
+fi
 
-# --- Python (uncomment if a Python-based generator is used) -------------------
-# if [ -f requirements.txt ]; then
-#   echo "[codex-setup] installing pip dependencies"
-#   pip install -r requirements.txt
-# fi
-
-echo "[codex-setup] done (no toolchain configured yet)"
+echo "[codex-setup] done"

.github/workflows/blank.yml

@@ -0,0 +1,62 @@
+name: Generate & Publish
+
+# Builds the static site from posts/ and publishes it to GitHub Pages.
+#
+# Triggers:
+#   - daily on a schedule (so AI-generated posts go live automatically)
+#   - on push to main (the production branch)
+#   - manually via the Actions tab
+#
+# Note: develop is the integration branch; publishing happens from main.
+# Merge develop -> main to release.
+
+on:
+  schedule:
+    # 23:17 UTC daily (~07:17 Asia/Shanghai). Off the :00 mark on purpose.
+    - cron: "17 23 * * *"
+  push:
+    branches: ["main"]
+  workflow_dispatch:
+
+permissions:
+  contents: read
+  pages: write
+  id-token: write
+
+# Allow only one concurrent deployment.
+concurrency:
+  group: "pages"
+  cancel-in-progress: true
+
+jobs:
+  build:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - uses: actions/setup-node@v4
+        with:
+          node-version: "20"
+          cache: "npm"
+
+      - name: Install dependencies
+        run: npm ci
+
+      - name: Build site
+        run: npm run build
+
+      - name: Upload Pages artifact
+        uses: actions/upload-pages-artifact@v3
+        with:
+          path: public
+
+  deploy:
+    needs: build
+    runs-on: ubuntu-latest
+    environment:
+      name: github-pages
+      url: ${{ steps.deployment.outputs.page_url }}
+    steps:
+      - name: Deploy to GitHub Pages
+        id: deployment
+        uses: actions/deploy-pages@v4

.github/workflows/publish.yml

@@ -0,0 +1,9 @@
+# Dependencies
+node_modules/
+
+# Build output (regenerated by scripts/build.mjs)
+public/
+
+# Logs / OS cruft
+*.log
+.DS_Store

.gitignore

@@ -4,61 +4,84 @@ Guidance for AI coding agents (OpenAI Codex and similar) working in the **OpenBl
 
 ## What this repo is
 
-OpenBlog is a personal blog project. The repository is currently an early
-scaffold — at the time of writing it contains only a `README.md` and a starter
-GitHub Actions workflow (`.github/workflows/blank.yml`). There is **no
-application code, build system, or test suite yet**. Treat the structure below
-as the intended target; create files as needed when implementing a task.
+OpenBlog is a personal blog system positioned as an **AI-driven, auto-generate &
+auto-publish blog**: daily thoughts become your memory, daily posts become your
+IP. See [README.md](README.md) for the full positioning.
+
+The repo now ships a working **static-site generator**: Markdown posts in
+`posts/` are rendered to a static site in `public/` and published to GitHub
+Pages by CI.
 
 ## Repository layout
 
 ```
 .
 ├── AGENTS.md                  # this file
-├── README.md
+├── README.md                  # positioning + usage
+├── package.json               # scripts: build, serve
+├── posts/                     # Markdown articles (the content source)
+│   └── *.md                   # front matter: title, date, tags, summary
+├── scripts/
+│   └── build.mjs              # static-site generator (Markdown -> public/)
+├── public/                    # build output (git-ignored, regenerated)
 └── .github/
     └── workflows/
-        └── blank.yml          # placeholder CI (echoes hello world)
+        └── publish.yml        # scheduled generate + publish to Pages
 ```
 
-When you add real code, prefer a conventional static-site / blog layout
-(e.g. `src/`, `content/` or `posts/` for Markdown articles, `public/` or
-`dist/` for build output). Document any structure you introduce back in this
-file and the `README.md`.
-
-## Branches
+## Branch policy — IMPORTANT
 
-- `main` — default / production branch.
-- `develop` — integration branch.
+- **`develop`** — integration branch. **This is the default branch for all work
+  and pushes.** Commit here and `git push origin develop`.
+- **`main`** — production / published branch. The publish workflow deploys from
+  `main`. Only merge `develop -> main` when explicitly releasing/publishing.
 
-Open pull requests against `main` unless the task says otherwise. Never force-push
-shared branches.
+**Default to `develop`.** Do NOT push or open PRs against `main` unless the task
+explicitly says to release/publish. Never force-push shared branches.
 
 ## Build, run, and test
 
-There is no build or test tooling configured yet. **Do not invent commands that
-do not exist.** If a task requires building or testing:
+Toolchain: Node.js (ESM), deps `marked` + `gray-matter`.
+
+```bash
+npm ci          # install (CI) — or npm install locally
+npm run build   # render posts/ -> public/
+npm run serve   # build, then serve public/ at http://localhost:8080
+```
+
+Validation for a change: `npm run build` succeeds and produces
+`public/index.html` plus one `public/posts/<slug>.html` per post. Verify the
+build runs before claiming it works.
 
-1. Choose a stack appropriate to a blog (e.g. a static-site generator), add its
-   manifest (`package.json`, etc.), and wire real `build`/`test` scripts.
-2. Update this section and the CI workflow with the actual commands you added.
-3. Verify commands run locally before claiming they work.
+## Authoring posts
+
+Add a `.md` file under `posts/` (filename becomes the URL slug). Front matter:
+
+```markdown
+---
+title: My thought
+date: 2026-06-11
+tags: [memory, ai]
+summary: One-line teaser shown on the index.
+---
+
+# My thought
+...
+```
 
-Until then, validation means: the repo still clones, Markdown/HTML renders, and
-the CI workflow stays green.
+All fields are optional — missing title falls back to the first heading or the
+filename; missing date falls back to file mtime.
 
 ## Conventions
 
 - Keep changes focused and minimal; match the style of surrounding files.
-- Blog content should be authored in Markdown.
-- Commit messages: short imperative subject line; explain the "why" in the body
-  when non-obvious.
-- Do not commit secrets, build artifacts, or large binaries.
+- Blog content is authored in Markdown.
+- Don't commit `node_modules/` or `public/` (both git-ignored).
+- Commit messages: short imperative subject; explain the "why" when non-obvious.
 
 ## For the Codex cloud agent
 
-- Read this file first to understand scope before making changes.
-- An optional environment setup script lives at `.codex/setup.sh` — run it to
-  prepare a working environment. It is a no-op until a real toolchain is added.
-- Make the smallest change that satisfies the task, open a PR against `main`,
-  and describe what you did and how you verified it.
+- Read this file first; respect the **develop-default** branch policy above.
+- Environment setup script: [.codex/setup.sh](.codex/setup.sh) (runs `npm ci`).
+- Make the smallest change that satisfies the task, push to `develop`, and
+  describe what you did and how you verified it (`npm run build`).

AGENTS.md

@@ -60,27 +60,59 @@ OpenBlog 用 AI Agent 把这条链路自动化:
 
 ## 项目状态
 
-> ⚠️ **早期阶段(Scaffold)**
+> 🚧 **早期可用(Early / MVP)**
 >
-> 当前仓库处于初始搭建阶段,尚未包含完整的应用代码与生成/发布流水线。
-> 上面描述的是项目的**目标定位与设计蓝图**,功能正在逐步落地。
+> 已具备一条可运行的静态站点流水线:`posts/` 里的 Markdown 文章会被生成器渲染成站点,并由 CI 定时构建、自动发布到 GitHub Pages。AI 自动撰写文章的环节正在逐步接入。
 
 ### 路线图(Roadmap)
 
+- [x] 静态站点生成器(Markdown → HTML)
+- [x] 自动构建与发布(GitHub Pages,定时 + 触发)
 - [ ] 记忆采集与结构化存储
-- [ ] AI 文章生成流水线
-- [ ] 自动构建与发布
-- [ ] 定时 / 触发式自动运行
+- [ ] AI 文章生成流水线(由记忆自动撰写)
 - [ ] 站点主题与个人化配置
 
 ## 快速开始
 
 ```bash
 git clone https://github.com/xiami303/OpenBlog.git
 cd OpenBlog
+npm install        # 安装依赖
+npm run build      # 渲染 posts/ → public/
+npm run serve      # 构建并本地预览 http://localhost:8080
 ```
 
-> 构建与运行命令将随功能落地补充。当前仓库尚无构建工具链 —— 详见 [AGENTS.md](AGENTS.md)。
+### 写一篇文章
+
+在 `posts/` 下新建一个 `.md` 文件(文件名即文章 URL slug),开头写好 front matter:
+
+```markdown
+---
+title: 我的想法
+date: 2026-06-11
+tags: [memory, ai]
+summary: 显示在首页的一句话简介。
+---
+
+# 我的想法
+...
+```
+
+所有字段都可省略 —— 没有 title 就用第一个标题或文件名,没有 date 就用文件修改时间。运行 `npm run build` 后,文章就会出现在站点上。
+
+## 目录结构
+
+```
+posts/                # Markdown 文章(内容源)
+scripts/build.mjs     # 静态站点生成器
+public/               # 构建产物(自动生成,已 gitignore)
+.github/workflows/    # 定时生成 + 自动发布到 Pages
+```
+
+## 分支规则
+
+- **`develop`** —— 集成分支,**日常开发与提交的默认分支**。
+- **`main`** —— 生产 / 已发布分支,CI 从这里部署到 Pages。仅在发布时把 `develop` 合并到 `main`。
 
 ## 参与开发