Update skills and add conda package recipe with CI/CD support

.gitattributes

@@ -0,0 +1,2 @@
+*.sh text eol=lf
+

.github/workflows/publish-conda.yml

@@ -0,0 +1,48 @@
+name: Publish Conda Package
+
+on:
+  push:
+    tags:
+      - 'v*'
+  workflow_dispatch:
+
+jobs:
+  build:
+    name: ${{ matrix.os }}
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - os: ubuntu-latest
+          - os: macos-13        # Intel Mac (osx-64)
+          - os: macos-latest    # Apple Silicon Mac (osx-arm64)
+
+    runs-on: ${{ matrix.os }}
+
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Setup Miniconda
+        uses: conda-incubator/setup-miniconda@v3
+        with:
+          auto-update-conda: true
+          channels: conda-forge
+          python-version: '3.11'
+
+      - name: Set package version from tag
+        run: |
+          TAG="${{ github.ref_name }}"
+          VERSION="${TAG#v}"
+          echo "VERSION=${VERSION}" >> $GITHUB_ENV
+          echo "Building gpumdkit ${VERSION}"
+
+      - name: Build conda package
+        run: conda build conda/ -c conda-forge
+
+      - name: Upload to Anaconda.org
+        if: startsWith(github.ref, 'refs/tags/v')
+        env:
+          ANACONDA_API_TOKEN: ${{ secrets.ANACONDA_TOKEN }}
+        run: |
+          anaconda upload --user ${{ vars.ANACONDA_USER }} \
+            $CONDA_PREFIX/conda-bld/*/gpumdkit-*.conda

AGENTS.md

@@ -1,38 +1,38 @@
 # GPUMDkit
 
-GPUMDkit is a Bash + Python command-line toolkit for [GPUMD](https://gpumd.org/) and [NEP](https://gitlab.com/brucefan1983/nep) — molecular dynamics and neuroevolution potential programs. It provides format conversion, structure sampling, workflows, property calculation, plotting, and analysis utilities.
+GPUMDkit is a Bash + Python CLI toolkit for GPUMD and NEP package: data conversion/sampling, workflows, calculations, analysis, plotting, simulation, training, and post-processing. Docs: <https://gpumdkit.cn/html/index.html>.
 
-**Entry point**: `gpumdkit.sh` — dual-mode (interactive menu + CLI flags):
+## Architecture
 
-```bash
-gpumdkit.sh                                # Interactive mode
-gpumdkit.sh -h                             # Show all CLI flags
-gpumdkit.sh -plt train                     # Plot NEP training
-gpumdkit.sh -calc msd dump.xyz Li 10       # Calculate MSD
-gpumdkit.sh -dp2xyz database train.xyz     # DeepMD npy → extxyz
+```text
+gpumdkit.sh -> src/fN_*.sh menu wrappers -> Scripts/<module>/ implementations
 ```
 
-**Documentation**: https://zhyan0603.github.io/GPUMDkit/
+- `gpumdkit.sh`: interactive and CLI entry point (`-h` lists commands; `-skill` locates skills).
+- `Scripts/`: conversion, sampling, workflow, calculator, analyzer, plot, and utility implementations.
+- `docs/tutorials/{en,zh}/`: bilingual sources; `docs/mkdocs.yml`: site config.
+- `skills/gpumdkit-skill/`: canonical agent skill; `gpumdkit-skill-zh/`: Chinese version.
 
-## Agent Skills
+## Agent Routing
 
-When the user asks about GPUMDkit features, use the skills in `.opencode/skills/`:
+For any GPUMDkit, GPUMD, NEP, atomistic-data, workflow, plotting, or post-processing task:
 
-| Skill | Use When |
-|-------|----------|
-| `gpumdkit-main` | General usage, navigation, entry point |
-| `gpumdkit-format-conversion` | Converting structure files between formats |
-| `gpumdkit-calculators` | Computing material properties |
-| `gpumdkit-analyzers` | Analyzing structures and datasets |
-| `gpumdkit-visualization` | Plotting simulation data |
-| `gpumdkit-workflows` | Batch processing and automation |
-| `gpumdkit-sampling` | Sampling and selecting structures |
+1. Read `skills/gpumdkit-skill/SKILL.md`.
+2. Classify the request and load only its routed references; combine them for cross-module work.
+3. Inspect local help/files, then prefer documented `gpumdkit.sh` commands over direct scripts.
 
-## If You Need to Modify Code
+Key references under `skills/gpumdkit-skill/references/`:
 
-**Read [skills/gpumdkit-contributing/SKILL.md](skills/gpumdkit-contributing/SKILL.md) first.** It contains all coding conventions, function templates, shell/Python patterns, and explicit rejections (things NOT to propose). The project follows a strict minimum-invasion principle — do not touch working code unless fixing a confirmed bug.
+- Navigation: `overview.md`.
+- Data/tools: `format-conversion.md`, `sampling.md`, `workflows.md`, `calculators.md`, `analyzers.md`.
+- Plots: `visualization.md`, `plotting-style.md`.
+- GPUMD: start with `gpumd.md`; it routes to `gpumd-{inputs,setup,ensembles,computes,outputs}.md`.
+- NEP: start with `nep.md`; it routes to `nep-{data,parameters,outputs}.md`.
+- Diffusion/conductivity fits: `arrhenius.md` plus relevant simulation, calculator, and plot references.
+- Code, docs, CLI, scripts, or skills: `contributing.md` plus the affected module reference.
 
-Key facts:
-- Tutorials are bilingual (`docs/tutorials/en/` + `zh/`). Feature changes must update both languages.
-- After Python edits: `python3 -m py_compile <file>` then clean `__pycache__`.
-- After doc edits: `mkdocs build -f docs/mkdocs.yml`.
+## Rules
+
+- Never invent consequential scientific or execution choices; ask the user.
+- Simulations, training, DFT, scheduler jobs, destructive operations, and expensive runs require explicit authorization.
+- Preserve unrelated changes. Before repository edits, read `contributing.md`; keep changes minimal, update both documentation languages for user-visible features, and run its validation checklist.

CONTRIBUTING.md

@@ -148,7 +148,7 @@ The structure of `GPUMDkit` consists of:
   - `f5_analyzers.sh`
   - `f6_plots.sh`
   - `f7_utilities.sh`
-- **`skills/`**: AI agent skill definitions (SKILL.md files for opencode/Claude Code integration)
+- **`skills/`**: Portable Agent Skills definition (`gpumdkit-skill/SKILL.md`) and on-demand references
 - **`docs/`**: Documentation files
   - `tutorials/en/` and `tutorials/zh/`: Bilingual tutorial pages
   - `mkdocs.yml`: MkDocs configuration for building tutorial HTML
@@ -185,7 +185,7 @@ To add a new feature accessible through the interactive menu:
    echo " Input <required_param1> <required_param2>"
    echo " Example: input.xyz output.lmp"
    echo " ------------>>"
-   read -r -a converter_args
+   read_menu_array converter_args || return 1
    echo " ---------------------------------------------------"
    python ${GPUMDkit_path}/Scripts/format_conversion/new_converter.py "${converter_args[@]}"
    echo " Code path: ${GPUMDkit_path}/Scripts/format_conversion/new_converter.py"
@@ -230,18 +230,28 @@ To add a new command-line flag or subcommand:
    touch Scripts/analyzer/analyze_bonds.py
    ```
 
-2. **Implement your functionality** with clear parameter requirements:
+2. **Implement your functionality** with clear parameter requirements. Python scripts
+   should own detailed argument parsing, help text, type conversion, file checks, and
+   user-facing error messages:
    ```python
    # Example: Scripts/analyzer/analyze_bonds.py
    import sys
 
-   if len(sys.argv) != 3:
-       print("Usage: gpumdkit.sh -analyze_bonds <input.xyz> <cutoff_distance>")
-       print("Example: gpumdkit.sh -analyze_bonds structure.xyz 3.0")
-       sys.exit(1)
+   args = sys.argv[1:]
+   if len(args) < 2 or args[0] in ("-h", "--help"):
+       print(" Usage: gpumdkit.sh -analyze_bonds <input.xyz> <cutoff_distance>")
+       print("    or: python3 analyze_bonds.py <input.xyz> <cutoff_distance>")
+       print("")
+       print(" Arguments:")
+       print("   input.xyz        Input extxyz file")
+       print("   cutoff_distance  Bond cutoff distance")
+       print("")
+       print(" Example: gpumdkit.sh -analyze_bonds structure.xyz 3.0")
+       print("")
+       sys.exit(0 if args and args[0] in ("-h", "--help") else 1)
 
-   input_file = sys.argv[1]
-   cutoff = float(sys.argv[2])
+   input_file = args[0]
+   cutoff = float(args[1])
 
    # Your implementation here
    ```
@@ -250,21 +260,17 @@ To add a new command-line flag or subcommand:
    > You may use a `main()` function with `if __name__ == "__main__":` guard if you prefer,
    > but it is not required. Most existing scripts execute at module level.
 
-3. **Add the command-line flag handler** in `gpumdkit.sh`:
+3. **Add the command-line flag handler** in `gpumdkit.sh`. Keep this layer as a
+   lightweight router. Detailed parameter validation should stay in the Python script,
+   not in the main shell entry point.
    ```bash
    # Find the command-line parsing section (the large "case $1 in" block)
    # Add your new flag in the appropriate location
 
    case $1 in
        # ... existing cases ...
        -analyze_bonds)
-           if [ ! -z "$2" ] && [ "$2" != "-h" ] && [ ! -z "$3" ]; then
-               python ${analyzer_path}/analyze_bonds.py $2 $3
-           else
-               echo " Usage: gpumdkit.sh -analyze_bonds <input.xyz> <cutoff_distance>"
-               echo " Example: gpumdkit.sh -analyze_bonds structure.xyz 3.0"
-               echo " Code path: ${analyzer_path}/analyze_bonds.py"
-           fi ;;
+           run_python_script "Your Name (your@email.com)" "${analyzer_path}/analyze_bonds.py" "${@:2}" ;;
        # ... rest of cases ...
    esac
    ```
@@ -316,7 +322,7 @@ esac
 
 - **Shell Scripts**:
   - Use `${variable}` for variable expansion
-  - Use `[ ! -z "$var" ]` for non-empty checks (project convention)
+  - Use `[ ! -z "$var" ]` for shell control flow and broad dispatch checks (project convention)
   - Interactive functions should follow the banner format:
     ```bash
     echo " >-------------------------------------------------<"
@@ -325,7 +331,7 @@ esac
     echo " | Developer: <Name> (<email>)                     |"
     echo " >-------------------------------------------------<"
     ```
-  - Use `read -r -a varname` for multi-word input, then pass with `"${varname[@]}"`
+  - Use `read_menu_choice varname || return 1` or `read_menu_array varname || return 1` for interactive input, then pass arrays with `"${varname[@]}"`
   - Shell scripts in `src/` should start with a file header block:
     ```bash
     # ============================================================
@@ -338,8 +344,9 @@ esac
 - **Python Scripts**:
   - Write clear, maintainable code
   - Use meaningful variable names
-  - Use `sys.argv` for argument parsing (consistent with most existing scripts)
-  - Provide clear usage messages with `print("Usage: ...")` and `print("Example: ...")`
+  - Use `args = sys.argv[1:]` for simple positional argument parsing, or `argparse` for complex option sets
+  - Provide clear usage messages with leading-space prints such as `print(" Usage: ...")` and `print(" Example: ...")`
+  - Put help and missing-argument checks before heavy optional imports so `-h` works without optional packages installed
   - If your script uses heavy/special packages (`NepTrain`, `calorine`, `dpdata`), add a
     `print_dependency_notice()` function to inform users about citation recommendations
 

README.md

@@ -6,10 +6,10 @@
    · 
   <a href="./docs/README_zh-CN.md">简体中文</a>
   &nbsp;·&nbsp;
-  <a href="https://zhyan0603.github.io/GPUMDkit/">Website</a> &nbsp;·&nbsp;
-  <a href="https://zhyan0603.github.io/GPUMDkit/htmls/index.html">Documentation</a>
+  <a href="https://gpumdkit.cn/">Website</a> &nbsp;·&nbsp;
+  <a href="https://gpumdkit.cn/htmls/index.html">Documentation</a>
   &nbsp;·&nbsp;
-  <a href="https://zhyan0603.github.io/GPUMDkit/gallery.html">Gallery</a>
+  <a href="https://gpumdkit.cn/gallery.html">Gallery</a>
   &nbsp;
 </p>
 <p align="center">
@@ -102,7 +102,7 @@ There are two options, <u>*interactive mode*</u> and <u>*command-line mode*</u>
              | |_| |  __/| |_| | |  | | |_| |   <| | |_
               \____|_|    \___/|_|  |_|____/|_|\_\_|\__|
 
-              GPUMDkit Version 1.5.6 (dev) (2026-06-17)
+              GPUMDkit Version 1.5.6 (dev) (2026-07-10)
         Core Developer: Zihan YAN (yanzihan@westlake.edu.cn)
      Main Contributors: Denan LI, Xin WU, Zhoulin LIU & Chen HUA
 
@@ -132,13 +132,14 @@ the help information:
 
 ```
 +-------------------------------------------------------------------------------------------------------+
-|                          GPUMDkit 1.5.6 (dev) (2026-06-17) Command Help                               |
+|                          GPUMDkit 1.5.6 (dev) (2026-07-10) Command Help                               |
 +-------------------------------------------------------------------------------------------------------+
 |                                          MAIN FUNCTIONS                                               |
 +-------------------------------------------------------------------------------------------------------+
 | -h            Show this help table            | -plt <type>        Plot and visualization tools       |
 | -calc <type>  Calculator tools                | -time <gpumd|nep>  Time-consuming analyzer            |
 | -update       Update GPUMDkit                 | -clean             Clean extra files in current dir   |
+| -skill        Show GPUMDkit agent skill info  | -doctor            Check Python environment           |
 +-------------------------------------------------------------------------------------------------------+
 |                                         FORMAT CONVERSION                                             |
 +-------------------------------------------------------------------------------------------------------+
@@ -150,7 +151,7 @@ the help information:
 | -traj2exyz    ASE traj -> extxyz              | -replicate         Replicate structure                |
 | -addgroup     Add group labels                | -addweight         Add structure weight in extxyz     |
 | -clean_xyz    Clean extra info in extxyz      | -get_frame         Extract specific frame             |
-| -frame_range  Extract frames by range         |                                                       |
+| -frame_range  Extract frames by range         | -dp2xyz            DeepMD npy -> extxyz               |
 +-------------------------------------------------------------------------------------------------------+
 |                                            ANALYSIS                                                   |
 +-------------------------------------------------------------------------------------------------------+
@@ -161,7 +162,7 @@ the help information:
 | -pda          Probability density analysis    | -filter_box        Filter by box-edge length          |
 | -pynep        Deprecated PyNEP sampling       | -nep_modifier      Modify NEP model interactively     |
 +-------------------------------------------------------------------------------------------------------+
-| Detailed usage: gpumdkit.sh -<option> -h    Plot details: gpumdkit.sh -plt <type> -h                  |
+| Python option help: gpumdkit.sh -<option> -h    Plot list: gpumdkit.sh -plt -h                     |
 +-------------------------------------------------------------------------------------------------------+
 ```
 
@@ -175,9 +176,9 @@ the help information:
 
 ```
  +-----------------------------------------------------------------------------------------------+
- |                     GPUMDkit 1.5.6 (dev) (2026-06-17) PLOT & VISUALIZATION TOOLS              |
+ |                     GPUMDkit 1.5.6 (dev) (2026-07-10) PLOT & VISUALIZATION TOOLS              |
  +-----------------------------------------------------------------------------------------------+
- |  Usage: gpumdkit.sh -plt <type>                        Help: gpumdkit.sh -plt <type> -h       |
+ |  Usage: gpumdkit.sh -plt <type>                        List: gpumdkit.sh -plt -h              |
  +-----------------------------------------------------------------------------------------------+
  |                                    NEP Training & Evaluation                                  |
  +-----------------------------------------------------------------------------------------------+
@@ -265,13 +266,13 @@ You can also save images as PNG if your device doesn't support visualization:
 gpumdkit.sh -plt thermo save
 ```
 
-Refer to our [documentation](https://zhyan0603.github.io/GPUMDkit/) for more detailed examples and command options.
+Refer to our [documentation](https://gpumdkit.cn/) for more detailed examples and command options.
 
 #### Custom Commands
 
 `GPUMDkit` now supports custom commands via `~/.gpumdkit.in`.
 
-You can add your own shortcuts (e.g., `gpumdkit.sh -yourcommand`) by defining functions in this file. This allows you to extend `GPUMDkit` with personal scripts. See [custom command documentation](https://zhyan0603.github.io/GPUMDkit/) for details.
+You can add your own shortcuts (e.g., `gpumdkit.sh -yourcommand`) by defining functions in this file. This allows you to extend `GPUMDkit` with personal scripts. See [custom command documentation](https://gpumdkit.cn/) for details.
 
 #### Tab Completion Support
 

Scripts/README.md

@@ -84,6 +84,7 @@ Structure sampling and selection tools:
 - Farthest point sampling (FPS)
 - Structure perturbation for training sets
 - Maximum force deviation selection
+- Training/test splitting with uniform, random, or FPS selection
 
 **Quick Start:**
 ```bash
@@ -118,11 +119,15 @@ Workflow automation scripts for high-throughput calculations:
 # Via interactive mode:
 gpumdkit.sh
 # Select: 3) Workflow
- ------------>>
- 301) SCF batch pretreatment
- 302) MD sample batch pretreatment (gpumd)
- 303) MD sample batch pretreatment (lmp)
- 000) Return to the main menu
+ +---------------------------------------------------------+
+ |                      WORKFLOW TOOLS                     |
+ +---------------------------------------------------------+
+ | 301) SCF batch pretreatment                             |
+ | 302) MD sample batch pretreatment (gpumd)               |
+ | 303) MD sample batch pretreatment (lmp)                 |
+ +---------------------------------------------------------+
+ | 000) Return to the main menu                            |
+ +---------------------------------------------------------+
 ```
 
 ## Usage Modes
@@ -134,23 +139,13 @@ gpumdkit.sh
 ```
 Navigate through numbered menus to access different functionalities.
 
-```
-         ____ ____  _   _ __  __ ____  _    _ _
-        / ___|  _ \| | | |  \/  |  _ \| | _(_) |_
-       | |  _| |_) | | | | |\/| | | | | |/ / | __|
-       | |_| |  __/| |_| | |  | | |_| |   <| | |_
-        \____|_|    \___/|_|  |_|____/|_|\_\_|\__|
-
-        GPUMDkit Version 1.5.6 (dev) (2026-06-17)
-  Core Developer: Zihan YAN (yanzihan@westlake.edu.cn)
- Main Contributors: Denan LI, Xin WU, Zhoulin LIU & Chen HUA
-
- ----------------------- GPUMD -----------------------
-  1) Format Conversion          2) Sample Structures
-  3) Workflow                   4) Calculators
-  5) Analyzer                   6) Visualization
-  7) Utilities                  8) Help                
-  0) Exit
+```text
+ ---------------------- GPUMD ------------------------
+ 1) Format Conversion          2) Sample Structures
+ 3) Workflow                   4) Calculators
+ 5) Analyzer                   6) Visualization
+ 7) Utilities                  8) Help
+ 0) Exit
  ------------>>
  Input the function number:
 ```
@@ -194,7 +189,7 @@ Check individual subdirectory READMEs for specific requirements.
 
 ## Support
 
-- **Documentation**: [https://zhyan0603.github.io/GPUMDkit/home.html](https://zhyan0603.github.io/GPUMDkit/home.html)
+- **Documentation**: [https://gpumdkit.cn/home.html](https://gpumdkit.cn/home.html)
 - **Issues**: [GitHub Issues](https://github.com/zhyan0603/GPUMDkit/issues)
 - **Contact**: Zihan YAN (yanzihan@westlake.edu.cn)