Language routing / 语言路由 Detect the user's language from the latest message.

• If the user writes in English, continue with this English file and respond in English.
• If the user writes in Chinese, read SKILL.md in this same directory and respond in Chinese. Always match the response language to the user's input language.

chatmosp-kmc-simulator

1. Core Responsibilities

1. Check and manage Wine environment (KMC engine dependency)
2. Prepare KMC input files (copy template from MOSP_database + user-specified steps/T/pp)
3. Execute KMC simulation and monitor progress
4. Warn on large step counts (≥ 20M steps)
5. Generate TOF / coverage plots and CSV data
6. Auto-regenerate plots if missing or on user request

2. Prerequisites

• ✅ MSR task completed (if starting from MSR cluster)
• ✅ parameter-builder has built KMC parameters
• ✅ User has confirmed via the 5-option prompt (see parameter-builder)
• ✅ Wine environment installed (check on first run, prompt install if missing)
• ✅ MSR-generated ini.xyz copied to KMC task directory (if applicable)
• ✅ External mosp-for-chatMOSP installed, with kmc_standalone.py, MOSP_database/, and engine/main.exe
• ❌ DO NOT bypass parameter-builder and build KMC parameters manually
• ❌ DO NOT reuse MSR's input.json — KMC must independently prepare complete parameters

3. Wine Environment

The KMC engine is the Windows main.exe, requiring Wine to run.

3.1 Check Wine

which wine64 || which wine

3.2 Install Wine (Ubuntu/Debian)

sudo dpkg --add-architecture i386
sudo apt-get update
sudo apt-get install wine64 wine32

3.3 Automatic environment check

• ✅ Wine installed → proceed normally
• ⚠️ Wine missing → prompt install instructions
• ❌ Version incompatible → prompt upgrade

4. Step-Count Warning

When KMC steps ≥ 20M, MUST warn user before execution:

⚠️ Calculation Time Warning:
Current KMC steps: {N}, estimated time: ~{estimated_hours} hours.
(Reference: 20M steps ≈ 12 hours, 40M steps ≈ 24+ hours)
Continue with KMC simulation?

Note: Running multiple conditions (e.g., multiple temperatures) requires separate calculations; total time multiplies.

5. Input Contract (Required fields in input.json)

Top-level fields

KMC object required fields

⚠️ CRITICAL: Gas entropy s1.S_gas / s2.S_gas MUST match MSR's Gas1_S / Gas2_S (use the same calculation formula).

6. Directory Structure

KMC task directory must be under MSR directory:

mosp-for-chatMOSP/OUTPUT/{msr_task_name}/
└── KMC_{steps}steps/             ← KMC task directory
    ├── input.json                ← outside INPUT/
    ├── ini.xyz                   ← copied from MSR
    ├── coverage.png              ← generated after run
    ├── coverage_steps.png        ← generated after run
    ├── tof.png                   ← generated after run
    ├── tof_time.png              ← generated after run
    ├── INPUT/                    ← should be empty before run
    └── OUTPUT/                   ← populated after run
        ├── rec_cov.data
        ├── rec_event.data
        └── rec_site_spc.data

⚠️ IMPORTANT:

• ini.xyz and input.json MUST be outside INPUT/ (kmc_standalone.py clears INPUT/OUTPUT)
• Ensure INPUT/ and OUTPUT/ are empty before running

7. Execution Steps

Step 1: Create KMC task directory

mkdir -p OUTPUT/Pd_CO9_O18_473K_101325Pa_R50/KMC_{steps}steps/INPUT
mkdir -p OUTPUT/Pd_CO9_O18_473K_101325Pa_R50/KMC_{steps}steps/OUTPUT

Step 2: Copy structure file

cp OUTPUT/{msr_task_name}/ini.xyz OUTPUT/{msr_task_name}/KMC_{steps}steps/ini.xyz

Step 3: Prepare input.json

# Copy template from MOSP_database (do not create manually)
cp mosp-for-chatMOSP/MOSP_database/{metal}-{reaction}.json \
   OUTPUT/{msr_task_name}/KMC_{steps}steps/input.json

# Adjust fields:
# - nLoop: user-specified steps
# - T: user-specified temperature
# - gas_pp: user-specified partial pressures
# - record_int: recording interval
# - s1.S_gas / s2.S_gas: recalculated using parameter-builder §7.1 formula

Step 4: Show parameters for user confirmation

Use parameter-builder §KMC parameter display format (5 options).

Step 5: Check Wine + step warning

• Check Wine environment
• Warn when steps ≥ 20M

Step 6: Execute KMC

python3 {mosp_root}/kmc_standalone.py \
  --xyz OUTPUT/{task_dir}/ini.xyz \
  --json OUTPUT/{task_dir}/input.json \
  --out-dir {task_dir}

{mosp_root} is normally $HOME/.openclaw/workspace/mosp-for-chatMOSP or $OPENCLAW_WORKSPACE/mosp-for-chatMOSP. Confirm that this directory exists before running.

Note: pass just the task directory name to --out-dir (e.g. {msr_task_name}/KMC_{steps}steps). The script automatically places it under OUTPUT/; even if an OUTPUT/ prefix is accidentally included it is de-duplicated, so no OUTPUT/OUTPUT/ nesting is produced.

8. Output Files

8.1 Sending Images to User (Feishu)

⚠️ One image per call: Feishu message tool only supports 1 image per attachment. Sending more than one will silently drop the extras. After KMC completes, you MUST send the 4 images one at a time with numbered labels.

Send order and labels:

1. Image 1/4: Coverage vs Time — coverage.png
2. Image 2/4: Coverage vs Steps — coverage_steps.png
3. Image 3/4: TOF vs Time — tof.png
4. Image 4/4: TOF vs Steps — tof_time.png

Example per message:

message(action=send, message="Image 1/4: Coverage vs Time",
         attachments=[{filePath:"...coverage.png", type:"image"}])

9. Output Check & Auto-Regenerate

After KMC completes (especially when user asks "is the task done?"), MUST check and regenerate plots if needed:

KMC_TASK_DIR="KMC task directory"
KMC_OUTPUT="$KMC_TASK_DIR/OUTPUT"

# 1. Check data files
if [ ! -f "$KMC_OUTPUT/rec_cov.data" ] || \
   [ ! -f "$KMC_OUTPUT/rec_event.data" ] || \
   [ ! -f "$KMC_OUTPUT/rec_site_spc.data" ]; then
  echo "❌ Data files missing, KMC may not have completed"
  exit 1
fi

# 2. Check step-count consistency
EXPECTED=$(grep -E "^[0-9]+\s+! Num of steps" "$KMC_TASK_DIR/INPUT/input.txt" | awk '{print $1}')
ACTUAL=$(tail -n 1 "$KMC_OUTPUT/rec_event.data" | awk '{print $2}')

if [ "$EXPECTED" != "$ACTUAL" ]; then
  echo "❌ Expected $EXPECTED steps, got $ACTUAL, KMC did not complete"
  exit 1
fi

# 3. Check plots (4 images), regenerate if missing
if [ -f "$KMC_TASK_DIR/coverage.png" ] && \
   [ -f "$KMC_TASK_DIR/coverage_steps.png" ] && \
   [ -f "$KMC_TASK_DIR/tof.png" ] && \
   [ -f "$KMC_TASK_DIR/tof_time.png" ]; then
  echo "✅ Plots already exist"
else
  python3 ../../utils/plot_kmc_data.py "$KMC_OUTPUT"
  echo "✅ Plots regenerated"
fi

Plotting script: mosp-for-chatMOSP/utils/plot_kmc_data.py

10. Error Handling

11. Cross-Skill Handoff

• MSR → KMC: After MSR completes, ini.xyz is produced. This skill independently fetches complete KMC parameters from MOSP_database. Do NOT reuse MSR's input.json.
• Parameter modification: If user changes temperature, parameter-builder MUST recalculate gas entropy per §7.5.
• Long-task query: When user asks "is the task done?", run §9 check script, regenerate plots if needed.

12. Dependencies

• mosp-for-chatMOSP — KMC engine (cloned, includes main.exe)
• chatmosp-parameter-builder — parameter building and gas entropy
• chatmosp-file-organizer — directory creation
• chatmosp-input-coordinator — task entry point
• Wine — required for Windows main.exe

13. File Structure

chatmosp-kmc-simulator/
├── SKILL.md       # Chinese reference
└── SKILL_cn.md    # This file (English)

14. Example

User: Run KMC simulation for Pt CO oxidation at 850K, 150Pa, 20M steps

System: [Recognize as KMC → parameter-builder searches params →
        Show 5-option confirmation → User confirms → Check Wine installed →
        ⚠️ 20M steps ≈ 12 hours, warn user → User confirms again →
        Prepare input.json → Invoke kmc_standalone.py → Monitor progress →
        Check output → Generate 4 images → Send one-by-one → Display results]

ChatMOSP Companion Skills

This skill is part of the ChatMOSP skill set. For complete functionality, install the companion ChatMOSP skills from ClawHub when they are missing:

• chatmosp-input-coordinator
• chatmosp-parameter-builder
• chatmosp-msr-generator
• chatmosp-kmc-simulator
• chatmosp-literature-search
• chatmosp-file-organizer

If a required companion skill is unavailable, ask the user to install the missing ChatMOSP skill from ClawHub before continuing. Use the GitHub repository only as a fallback.