[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6G5RBYSTWZ4KZ7N6WRR5HVQ":3,"skill-same:01M6G5RBYSTWZ4KZ7N6WRR5HVQ":81},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":15,"categories":16,"keywords":22,"publishedAt":39,"linkOnly":40,"markdownContent":41,"markdownBody":42,"frontMatter":43,"refName":44,"commitSha":45,"skillDirectory":46,"forksCount":47,"repositoryHtmlUrl":48,"repositoryHomepageUrl":6,"ownerHtmlUrl":49,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":50,"install":51,"resources":64},"01M6G5RBYSTWZ4KZ7N6WRR5HVQ","turix-mac",null,"Computer Use Agent (CUA) for macOS automation using TuriX. Use when you need to perform visual tasks on the desktop, such as opening apps, clicking buttons, or navigating UIs that don't have a CLI or API.","kbarbel640-del\u002Fskills","kbarbel640-del","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F255315490?v=4","01M6CJPPQGCCESXMG3K7BBGQ5J","01M6CJPPX2PB2BCZERFVM2C2B3",1,"MIT","https:\u002F\u002Fgithub.com\u002Fkbarbel640-del\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Ftongyu-yan\u002Fturix-cua\u002FSKILL.md",[17],{"categoryNo":18,"categoryName":19,"taxonomyType":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000023","Skill与智能体开发","SCENARIO","使用场景",[23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38],"mac","turix mac","macos","cli","automation","tasks","desktop","Computer Use","api","computer","use","computer use agent","visual","as","apps","you","2026-11-10 10:12:30",false,"---\nname: turix-mac\ndescription: Computer Use Agent (CUA) for macOS automation using TuriX. Use when you need to perform visual tasks on the desktop, such as opening apps, clicking buttons, or navigating UIs that don't have a CLI or API.\n---\n\n# TuriX-Mac Skill\n\nThis skill allows Clawdbot to control the macOS desktop visually using the TuriX Computer Use Agent.\n\n## When to Use\n\n- When asked to perform actions on the Mac desktop (e.g., \"Open Spotify and play my liked songs\").\n- When navigating applications that lack command-line interfaces.\n- For multi-step visual workflows (e.g., \"Find the latest invoice in my email and upload it to the company portal\").\n- When you need the agent to plan, reason, and execute complex tasks autonomously.\n\n## Key Features\n\n### ? Multi-Model Architecture\nTuriX uses a sophisticated multi-model system:\n- **Brain**: Understands the task and generates step-by-step plans\n- **Actor**: Executes precise UI actions based on visual understanding\n- **Planner**: Coordinates high-level task decomposition (when `use_plan: true`)\n- **Memory**: Maintains context across task steps\n\n### ? Skills System\nSkills are markdown playbooks that guide the agent for specific domains:\n- `github-web-actions`: GitHub navigation, repo search, starring\n- `browser-tasks`: General web browser operations\n- Custom skills can be added to the `skills\u002F` directory\n\n### ? Resume Capability\nThe agent can resume interrupted tasks by setting a stable `agent_id`.\n\n## Running TuriX\n\n### Basic Task\n```bash\nskills\u002Flocal\u002Fturix-mac\u002Fscripts\u002Frun_turix.sh \"Open Chrome and go to github.com\"\n```\n\n### Resume Interrupted Task\n```bash\nskills\u002Flocal\u002Fturix-mac\u002Fscripts\u002Frun_turix.sh --resume my-task-001\n```\n\n> ✅ **Note**: `run_turix.sh` updates `examples\u002Fconfig.json` for you (task, resume, `use_plan`, `use_skills`). If you want to keep a hand-edited config, skip passing a task and edit `examples\u002Fconfig.json` directly.\n\n\n### Tips for Effective Tasks\n\n**✅ Good Examples:**\n- \"Open Safari, go to google.com, search for 'TuriX AI', and click the first result\"\n- \"Open System Settings, click on Dark Mode, then return to System Settings\"\n- \"Open Finder, navigate to Documents, and create a new folder named 'Project X'\"\n\n**❌ Avoid:**\n- Vague instructions: \"Help me\" or \"Fix this\"\n- Impossible actions: \"Delete all files\"\n- Tasks requiring system-level permissions without warning\n\n**? Best Practices:**\n1. Be specific about the target application\n2. Break complex tasks into clear steps, but do not mention the precise coordinates on the screen.\n\n## Hotkeys\n\n- **Force Stop**: `Cmd+Shift+2` - Immediately stops the agent\n\n## Monitoring & Logs\n\nLogs are saved to `.turix_tmp\u002Flogging.log` in the project directory. Check this for:\n- Step-by-step execution details\n- LLM interactions and reasoning\n- Errors and recovery attempts\n\n## Important Notes\n\n### How TuriX Runs\n- TuriX can be started via clawdbot `exec` with `pty:true` mode\n- The first launch takes 2-5 minutes to load all AI models (Brain, Actor, Planner, Memory)\n- Background output is buffered - you won't see live progress until task completes or stops\n\n### Before Running\n**Always set PATH first:**\n```bash\nexport PATH=\"\u002Fusr\u002Fsbin:$PATH\"\ncd your_dir\u002FTuriX-CUA\n\u002Fopt\u002Fanaconda3\u002Fenvs\u002Fturix_env\u002Fbin\u002Fpython examples\u002Fmain.py\n```\n\n**Why?** The `screencapture` tool is located at `\u002Fusr\u002Fsbin\u002Fscreencapture`, which is not in the default PATH.\n\n### Checking if TuriX is Running\n```bash\n# Check process\nps aux | grep \"python.*main\" | grep -v grep\n\n# Should show something like:\n# user  57425  0.0  2.4 412396704 600496 s143  Ss+  5:56PM   0:04.76 \u002Fopt\u002Fanaconda3\u002Fenvs\u002Fturix_env\u002Fbin\u002Fpython examples\u002Fmain.py\n```\n\n**Note:** The `.turix_tmp` directory may not be created until TuriX starts executing steps.\n\n## Troubleshooting\n\n### Common Issues\n\n| Error | Solution |\n|-------|----------|\n| `NoneType has no attribute 'save'` | Screen recording permission missing. Grant in System Settings and restart Terminal. |\n| `Screen recording access denied` | Run: `osascript -e 'tell application \"Safari\" to do JavaScript \"alert(1)\"'` and click Allow |\n| `Conda environment not found` | Ensure `turix_env` exists: `conda create -n turix_env python=3.12` |\n| Module import errors | Activate environment: `conda activate turix_env` then `pip install -r requirements.txt` |\n| Permission errors for keyboard listener | Add Terminal\u002FIDE to **Accessibility** permissions |\n\n### Debug Mode\n\nLogs include DEBUG level by default. Check:\n```bash\ntail -f your_dir\u002FTuriX-CUA\u002F.turix_tmp\u002Flogging.log\n```\n\n## Architecture\n\n```\nUser Request\n     ↓\n[Clawdbot] → [TuriX Skill] → [run_turix.sh] → [TuriX Agent]\n                                              ↓\n                    ┌─────────────────────────┼─────────────────────────┐\n                    ↓                         ↓                         ↓\n               [Planner]                 [Brain]                  [Memory]\n                    ↓                         ↓                         ↓\n                                         [Actor] ───→ [Controller] ───→ [macOS UI]\n```\n\n## Skill System Details\n\nSkills are markdown files with YAML frontmatter in the `skills\u002F` directory:\n\n```md\n---\nname: skill-name\ndescription: When to use this skill\n---\n# Skill Instructions\nHigh-level workflow like: Open Safari,then go to Google.\n```\n\nThe Planner selects relevant skills based on name\u002Fdescription; the Brain uses full content for step guidance.\n\n## Advanced Options\n\n| Option | Description |\n|--------|-------------|\n| `use_plan: true` | Enable planning for complex tasks |\n| `use_skills: true` | Enable skill selection |\n| `resume: true` | Resume from previous interruption |\n| `max_steps: N` | Limit total steps (default: 100) |\n| `max_actions_per_step: N` | Actions per step (default: 5) |\n| `force_stop_hotkey` | Custom hotkey to stop agent |\n\n---\n\n## TuriX Skills System\n\nTuriX supports **Skills**: markdown playbooks that help the agent behave more reliably in specific domains.\n\n### 1. Built-in Skills\n\n| Skill | Use |\n|-------|-----|\n| `github-web-actions` | GitHub web actions (search repos, star, etc.) |\n\n### 2. Create a Custom Skill\n\nCreate a `.md` file in the TuriX project's `skills\u002F` directory:\n\n```md\n---\nname: my-custom-skill\ndescription: When performing X specific task\n---\n# Custom Skill\n\n## Guidelines\n- Step 1: Do this first\n- Step 2: Then do that\n- Step 3: Verify the result\n```\n\n**Field definitions:**\n- `name`: Skill identifier (used by the Planner to select)\n- `description`: When to use this skill (Planner matches on this)\n- The body below: Full execution guide (used by the Brain)\n\n### 3. Enable Skills\n\nIn `examples\u002Fconfig.json`:\n\n```json\n{\n  \"agent\": {\n    \"use_plan\": true,\n    \"use_skills\": true,\n    \"skills_dir\": \"skills\",\n    \"skills_max_chars\": 4000\n  }\n}\n```\n\n### 4. Run a Task with Skills\n\n```bash\nskills\u002Flocal\u002Fturix-mac\u002Fscripts\u002Frun_turix.sh \"Search for turix-cua on GitHub and star it\"\n```\n\nThe agent will automatically:\n1. Planner reads the skill name and description\n2. Selects relevant skills\n3. Brain uses the full skill content to guide execution\n\n### 5. Chinese Text Support\n\n**Background:**\nPassing Chinese text through shell interpolation can mangle UTF-8, and interpolating untrusted text into a heredoc is unsafe.\n\n**Solution:**\nThe `run_turix.sh` script uses Python to handle UTF-8 correctly and reads task text from environment variables:\n\n```python\nimport json\n\n# Read with UTF-8\nwith open(config_path, 'r', encoding='utf-8') as f:\n    data = json.load(f)\n\n# Write without escaping non-ASCII text\nwith open(config_path, 'w', encoding='utf-8') as f:\n    json.dump(data, f, indent=2, ensure_ascii=False)\n```\n\n**Key points:**\n1. Always use `encoding='utf-8'` when reading\u002Fwriting files\n2. Use `ensure_ascii=False` to preserve non-ASCII text\n3. Pass task content via environment variables or stdin, and use a single-quoted heredoc to avoid shell interpolation\n\n### 6. Document Creation Best Practices\n\n**Challenges:**\n- Asking TuriX to collect news, then create and send a document directly\n- TuriX is a GUI agent, so it can be slow and less deterministic. Prefer using TuriX only for tasks Clawdbot cannot do or where TuriX is faster.\n\n**Recommended approach:** create the document yourself and let TuriX only send it\n1. Create the Word document with python-docx\n2. Let TuriX only send the file\n\n```python\nfrom docx import Document\ndoc = Document()\ndoc.add_heading('Title')\ndoc.save('\u002Fpath\u002Fto\u002Ffile.docx')\n```\n\n**Suggested workflow:**\n1. Use `web_fetch` to gather information\n2. Use Python to create the Word document\n3. Use TuriX to send the file. Specify the file path and say to send the file, not just the file name.\n4. If you really need TuriX to manually create a Word document and type in collected information, put the content in turix skills (for large amounts) or in the task name (for small amounts).\n\n### 7. Example: Add a New Skill\n\nCreate `skills\u002Fbrowser-tasks.md`:\n\n```md\n---\nname: browser-tasks\ndescription: When performing tasks in a web browser (search, navigate, fill forms).\n---\n# Browser Tasks\n\n## Navigation\n- Use the address bar or search box to navigate\n- Open new tabs for each distinct task\n- Wait for page to fully load before proceeding\n\n## Forms\n- Click on input fields to focus\n- Type content clearly\n- Look for submit\u002Fbutton to complete actions\n\n## Safety\n- Confirm before submitting forms\n- Do not download files without user permission\n```\n\n### 8. Skill Development Tips\n\n1. **Be precise in the description** - helps the Planner select correctly\n2. **Make steps clear** - the Brain needs explicit guidance\n3. **Include safety checks** - confirmations for important actions\n4. **Keep it concise** - recommended under 4000 characters\n\n---\n\n## Monitoring and Debugging Guide\n\n### 1. Run a Task\n\n```bash\n# Run in background (recommended)\ncd your_dir\u002Fclawd\u002Fskills\u002Flocal\u002Fturix-mac\u002Fscripts\n.\u002Frun_turix.sh \"Your task description\" --background\n\n# Or use timeout to set a max runtime\n.\u002Frun_turix.sh \"Task\" &\n```\n\n### 2. Monitor Progress\n\n**Method 1: Session logs**\n```bash\n# List running sessions\nclawdbot sessions_list\n\n# View history\nclawdbot sessions_history \u003Csession_key>\n```\n\n**Method 2: TuriX logs**\n```bash\n# Tail logs in real time\ntail -f your_dir\u002FTuriX-CUA\u002F.turix_tmp\u002Flogging.log\n\n# Or inspect completed step files\nls -lt your_dir\u002FTuriX-CUA\u002Fexamples\u002F.turix_tmp\u002Fbrain_llm_interactions.log_brain_*.txt\n```\n\n**Method 3: Check processes**\n```bash\nps aux | grep \"python.*main.py\" | grep -v grep\n```\n\n**Method 4: Check generated files**\n```bash\n# List files created by the agent\nls -la your_dir\u002FTuriX-CUA\u002Fexamples\u002F.turix_tmp\u002F*.txt\n```\n\n### 3. Log File Reference\n\n| File | Description |\n|------|-------------|\n| `logging.log` | Main log file |\n| `brain_llm_interactions.log_brain_N.txt` | Brain model conversations (one per step) |\n| `actor_llm_interactions.log_actor_N.txt` | Actor model conversations (one per step) |\n\n**Key log markers:**\n- `? Step N` - New step started\n- `✅ Eval: Success\u002FFailed` - Current step evaluation\n- `? Goal to achieve this step` - Current goal\n- `?️  Action` - Executed action\n- `✅ Task completed successfully` - Task completed\n\n### 4. Common Monitoring Issues\n\n| Issue | Check |\n|-------|-------|\n| Process unresponsive | `ps aux | grep main.py` |\n| Stuck on step 1 | Check whether `.turix_tmp\u002F` was created |\n| Model loading is slow | First run can take 1-2 minutes to load models |\n| No log output | Check `config.json` `logging_level` |\n\n### 5. Force Stop\n\n**Hotkey**: `Cmd+Shift+2` - stop the agent immediately\n\n**Command**:\n```bash\npkill -f \"python examples\u002Fmain.py\"\n```\n\n### 6. View Results\n\nAfter completion, the agent will:\n1. Create interaction logs in `.turix_tmp\u002F`\n2. Create record files (if `record_info` is used)\n3. Keep screenshots in memory for subsequent steps\n\n**Example: view a summary file**\n```bash\ncat your_dir\u002FTuriX-CUA\u002Fexamples\u002F.turix_tmp\u002Flatest_ai_news_summary_jan2026.txt\n```\n\n### 7. Debugging Tips\n\n1. **Inspect Brain reasoning**: check `brain_llm_interactions.log_brain_*.txt` for `analysis` and `next_goal`\n2. **Inspect Actor actions**: check `actor_llm_interactions.log_actor_*.txt` for actions\n3. **Check screenshots**: TuriX captures a screenshot each step (kept in memory)\n4. **Read record files**: the agent uses `record_info` to save key info to `.txt` files\n\n### 8. Example Monitoring Flow\n\n```bash\n# 1. Run a task\n.\u002Frun_turix.sh \"Search AI news and summarize\" &\n\n# 2. Wait a few seconds and check the process\nsleep 10 && ps aux | grep main.py\n\n# 3. Check if logs are being created\nls -la your_dir\u002FTuriX-CUA\u002Fexamples\u002F.turix_tmp\u002F\n\n# 4. Tail progress in real time\ntail -f your_dir\u002FTuriX-CUA\u002F.turix_tmp\u002Flogging.log\n\n# 5. Check current step count\nls your_dir\u002FTuriX-CUA\u002Fexamples\u002F.turix_tmp\u002Fbrain_llm_interactions.log_brain_*.txt | wc -l\n```\n","# TuriX-Mac Skill\n\nThis skill allows Clawdbot to control the macOS desktop visually using the TuriX Computer Use Agent.\n\n## When to Use\n\n- When asked to perform actions on the Mac desktop (e.g., \"Open Spotify and play my liked songs\").\n- When navigating applications that lack command-line interfaces.\n- For multi-step visual workflows (e.g., \"Find the latest invoice in my email and upload it to the company portal\").\n- When you need the agent to plan, reason, and execute complex tasks autonomously.\n\n## Key Features\n\n### ? Multi-Model Architecture\nTuriX uses a sophisticated multi-model system:\n- **Brain**: Understands the task and generates step-by-step plans\n- **Actor**: Executes precise UI actions based on visual understanding\n- **Planner**: Coordinates high-level task decomposition (when `use_plan: true`)\n- **Memory**: Maintains context across task steps\n\n### ? Skills System\nSkills are markdown playbooks that guide the agent for specific domains:\n- `github-web-actions`: GitHub navigation, repo search, starring\n- `browser-tasks`: General web browser operations\n- Custom skills can be added to the `skills\u002F` directory\n\n### ? Resume Capability\nThe agent can resume interrupted tasks by setting a stable `agent_id`.\n\n## Running TuriX\n\n### Basic Task\n```bash\nskills\u002Flocal\u002Fturix-mac\u002Fscripts\u002Frun_turix.sh \"Open Chrome and go to github.com\"\n```\n\n### Resume Interrupted Task\n```bash\nskills\u002Flocal\u002Fturix-mac\u002Fscripts\u002Frun_turix.sh --resume my-task-001\n```\n\n> ✅ **Note**: `run_turix.sh` updates `examples\u002Fconfig.json` for you (task, resume, `use_plan`, `use_skills`). If you want to keep a hand-edited config, skip passing a task and edit `examples\u002Fconfig.json` directly.\n\n\n### Tips for Effective Tasks\n\n**✅ Good Examples:**\n- \"Open Safari, go to google.com, search for 'TuriX AI', and click the first result\"\n- \"Open System Settings, click on Dark Mode, then return to System Settings\"\n- \"Open Finder, navigate to Documents, and create a new folder named 'Project X'\"\n\n**❌ Avoid:**\n- Vague instructions: \"Help me\" or \"Fix this\"\n- Impossible actions: \"Delete all files\"\n- Tasks requiring system-level permissions without warning\n\n**? Best Practices:**\n1. Be specific about the target application\n2. Break complex tasks into clear steps, but do not mention the precise coordinates on the screen.\n\n## Hotkeys\n\n- **Force Stop**: `Cmd+Shift+2` - Immediately stops the agent\n\n## Monitoring & Logs\n\nLogs are saved to `.turix_tmp\u002Flogging.log` in the project directory. Check this for:\n- Step-by-step execution details\n- LLM interactions and reasoning\n- Errors and recovery attempts\n\n## Important Notes\n\n### How TuriX Runs\n- TuriX can be started via clawdbot `exec` with `pty:true` mode\n- The first launch takes 2-5 minutes to load all AI models (Brain, Actor, Planner, Memory)\n- Background output is buffered - you won't see live progress until task completes or stops\n\n### Before Running\n**Always set PATH first:**\n```bash\nexport PATH=\"\u002Fusr\u002Fsbin:$PATH\"\ncd your_dir\u002FTuriX-CUA\n\u002Fopt\u002Fanaconda3\u002Fenvs\u002Fturix_env\u002Fbin\u002Fpython examples\u002Fmain.py\n```\n\n**Why?** The `screencapture` tool is located at `\u002Fusr\u002Fsbin\u002Fscreencapture`, which is not in the default PATH.\n\n### Checking if TuriX is Running\n```bash\n# Check process\nps aux | grep \"python.*main\" | grep -v grep\n\n# Should show something like:\n# user  57425  0.0  2.4 412396704 600496 s143  Ss+  5:56PM   0:04.76 \u002Fopt\u002Fanaconda3\u002Fenvs\u002Fturix_env\u002Fbin\u002Fpython examples\u002Fmain.py\n```\n\n**Note:** The `.turix_tmp` directory may not be created until TuriX starts executing steps.\n\n## Troubleshooting\n\n### Common Issues\n\n| Error | Solution |\n|-------|----------|\n| `NoneType has no attribute 'save'` | Screen recording permission missing. Grant in System Settings and restart Terminal. |\n| `Screen recording access denied` | Run: `osascript -e 'tell application \"Safari\" to do JavaScript \"alert(1)\"'` and click Allow |\n| `Conda environment not found` | Ensure `turix_env` exists: `conda create -n turix_env python=3.12` |\n| Module import errors | Activate environment: `conda activate turix_env` then `pip install -r requirements.txt` |\n| Permission errors for keyboard listener | Add Terminal\u002FIDE to **Accessibility** permissions |\n\n### Debug Mode\n\nLogs include DEBUG level by default. Check:\n```bash\ntail -f your_dir\u002FTuriX-CUA\u002F.turix_tmp\u002Flogging.log\n```\n\n## Architecture\n\n```\nUser Request\n     ↓\n[Clawdbot] → [TuriX Skill] → [run_turix.sh] → [TuriX Agent]\n                                              ↓\n                    ┌─────────────────────────┼─────────────────────────┐\n                    ↓                         ↓                         ↓\n               [Planner]                 [Brain]                  [Memory]\n                    ↓                         ↓                         ↓\n                                         [Actor] ───→ [Controller] ───→ [macOS UI]\n```\n\n## Skill System Details\n\nSkills are markdown files with YAML frontmatter in the `skills\u002F` directory:\n\n```md\n---\nname: skill-name\ndescription: When to use this skill\n---\n# Skill Instructions\nHigh-level workflow like: Open Safari,then go to Google.\n```\n\nThe Planner selects relevant skills based on name\u002Fdescription; the Brain uses full content for step guidance.\n\n## Advanced Options\n\n| Option | Description |\n|--------|-------------|\n| `use_plan: true` | Enable planning for complex tasks |\n| `use_skills: true` | Enable skill selection |\n| `resume: true` | Resume from previous interruption |\n| `max_steps: N` | Limit total steps (default: 100) |\n| `max_actions_per_step: N` | Actions per step (default: 5) |\n| `force_stop_hotkey` | Custom hotkey to stop agent |\n\n---\n\n## TuriX Skills System\n\nTuriX supports **Skills**: markdown playbooks that help the agent behave more reliably in specific domains.\n\n### 1. Built-in Skills\n\n| Skill | Use |\n|-------|-----|\n| `github-web-actions` | GitHub web actions (search repos, star, etc.) |\n\n### 2. Create a Custom Skill\n\nCreate a `.md` file in the TuriX project's `skills\u002F` directory:\n\n```md\n---\nname: my-custom-skill\ndescription: When performing X specific task\n---\n# Custom Skill\n\n## Guidelines\n- Step 1: Do this first\n- Step 2: Then do that\n- Step 3: Verify the result\n```\n\n**Field definitions:**\n- `name`: Skill identifier (used by the Planner to select)\n- `description`: When to use this skill (Planner matches on this)\n- The body below: Full execution guide (used by the Brain)\n\n### 3. Enable Skills\n\nIn `examples\u002Fconfig.json`:\n\n```json\n{\n  \"agent\": {\n    \"use_plan\": true,\n    \"use_skills\": true,\n    \"skills_dir\": \"skills\",\n    \"skills_max_chars\": 4000\n  }\n}\n```\n\n### 4. Run a Task with Skills\n\n```bash\nskills\u002Flocal\u002Fturix-mac\u002Fscripts\u002Frun_turix.sh \"Search for turix-cua on GitHub and star it\"\n```\n\nThe agent will automatically:\n1. Planner reads the skill name and description\n2. Selects relevant skills\n3. Brain uses the full skill content to guide execution\n\n### 5. Chinese Text Support\n\n**Background:**\nPassing Chinese text through shell interpolation can mangle UTF-8, and interpolating untrusted text into a heredoc is unsafe.\n\n**Solution:**\nThe `run_turix.sh` script uses Python to handle UTF-8 correctly and reads task text from environment variables:\n\n```python\nimport json\n\n# Read with UTF-8\nwith open(config_path, 'r', encoding='utf-8') as f:\n    data = json.load(f)\n\n# Write without escaping non-ASCII text\nwith open(config_path, 'w', encoding='utf-8') as f:\n    json.dump(data, f, indent=2, ensure_ascii=False)\n```\n\n**Key points:**\n1. Always use `encoding='utf-8'` when reading\u002Fwriting files\n2. Use `ensure_ascii=False` to preserve non-ASCII text\n3. Pass task content via environment variables or stdin, and use a single-quoted heredoc to avoid shell interpolation\n\n### 6. Document Creation Best Practices\n\n**Challenges:**\n- Asking TuriX to collect news, then create and send a document directly\n- TuriX is a GUI agent, so it can be slow and less deterministic. Prefer using TuriX only for tasks Clawdbot cannot do or where TuriX is faster.\n\n**Recommended approach:** create the document yourself and let TuriX only send it\n1. Create the Word document with python-docx\n2. Let TuriX only send the file\n\n```python\nfrom docx import Document\ndoc = Document()\ndoc.add_heading('Title')\ndoc.save('\u002Fpath\u002Fto\u002Ffile.docx')\n```\n\n**Suggested workflow:**\n1. Use `web_fetch` to gather information\n2. Use Python to create the Word document\n3. Use TuriX to send the file. Specify the file path and say to send the file, not just the file name.\n4. If you really need TuriX to manually create a Word document and type in collected information, put the content in turix skills (for large amounts) or in the task name (for small amounts).\n\n### 7. Example: Add a New Skill\n\nCreate `skills\u002Fbrowser-tasks.md`:\n\n```md\n---\nname: browser-tasks\ndescription: When performing tasks in a web browser (search, navigate, fill forms).\n---\n# Browser Tasks\n\n## Navigation\n- Use the address bar or search box to navigate\n- Open new tabs for each distinct task\n- Wait for page to fully load before proceeding\n\n## Forms\n- Click on input fields to focus\n- Type content clearly\n- Look for submit\u002Fbutton to complete actions\n\n## Safety\n- Confirm before submitting forms\n- Do not download files without user permission\n```\n\n### 8. Skill Development Tips\n\n1. **Be precise in the description** - helps the Planner select correctly\n2. **Make steps clear** - the Brain needs explicit guidance\n3. **Include safety checks** - confirmations for important actions\n4. **Keep it concise** - recommended under 4000 characters\n\n---\n\n## Monitoring and Debugging Guide\n\n### 1. Run a Task\n\n```bash\n# Run in background (recommended)\ncd your_dir\u002Fclawd\u002Fskills\u002Flocal\u002Fturix-mac\u002Fscripts\n.\u002Frun_turix.sh \"Your task description\" --background\n\n# Or use timeout to set a max runtime\n.\u002Frun_turix.sh \"Task\" &\n```\n\n### 2. Monitor Progress\n\n**Method 1: Session logs**\n```bash\n# List running sessions\nclawdbot sessions_list\n\n# View history\nclawdbot sessions_history \u003Csession_key>\n```\n\n**Method 2: TuriX logs**\n```bash\n# Tail logs in real time\ntail -f your_dir\u002FTuriX-CUA\u002F.turix_tmp\u002Flogging.log\n\n# Or inspect completed step files\nls -lt your_dir\u002FTuriX-CUA\u002Fexamples\u002F.turix_tmp\u002Fbrain_llm_interactions.log_brain_*.txt\n```\n\n**Method 3: Check processes**\n```bash\nps aux | grep \"python.*main.py\" | grep -v grep\n```\n\n**Method 4: Check generated files**\n```bash\n# List files created by the agent\nls -la your_dir\u002FTuriX-CUA\u002Fexamples\u002F.turix_tmp\u002F*.txt\n```\n\n### 3. Log File Reference\n\n| File | Description |\n|------|-------------|\n| `logging.log` | Main log file |\n| `brain_llm_interactions.log_brain_N.txt` | Brain model conversations (one per step) |\n| `actor_llm_interactions.log_actor_N.txt` | Actor model conversations (one per step) |\n\n**Key log markers:**\n- `? Step N` - New step started\n- `✅ Eval: Success\u002FFailed` - Current step evaluation\n- `? Goal to achieve this step` - Current goal\n- `?️  Action` - Executed action\n- `✅ Task completed successfully` - Task completed\n\n### 4. Common Monitoring Issues\n\n| Issue | Check |\n|-------|-------|\n| Process unresponsive | `ps aux | grep main.py` |\n| Stuck on step 1 | Check whether `.turix_tmp\u002F` was created |\n| Model loading is slow | First run can take 1-2 minutes to load models |\n| No log output | Check `config.json` `logging_level` |\n\n### 5. Force Stop\n\n**Hotkey**: `Cmd+Shift+2` - stop the agent immediately\n\n**Command**:\n```bash\npkill -f \"python examples\u002Fmain.py\"\n```\n\n### 6. View Results\n\nAfter completion, the agent will:\n1. Create interaction logs in `.turix_tmp\u002F`\n2. Create record files (if `record_info` is used)\n3. Keep screenshots in memory for subsequent steps\n\n**Example: view a summary file**\n```bash\ncat your_dir\u002FTuriX-CUA\u002Fexamples\u002F.turix_tmp\u002Flatest_ai_news_summary_jan2026.txt\n```\n\n### 7. Debugging Tips\n\n1. **Inspect Brain reasoning**: check `brain_llm_interactions.log_brain_*.txt` for `analysis` and `next_goal`\n2. **Inspect Actor actions**: check `actor_llm_interactions.log_actor_*.txt` for actions\n3. **Check screenshots**: TuriX captures a screenshot each step (kept in memory)\n4. **Read record files**: the agent uses `record_info` to save key info to `.txt` files\n\n### 8. Example Monitoring Flow\n\n```bash\n# 1. Run a task\n.\u002Frun_turix.sh \"Search AI news and summarize\" &\n\n# 2. Wait a few seconds and check the process\nsleep 10 && ps aux | grep main.py\n\n# 3. Check if logs are being created\nls -la your_dir\u002FTuriX-CUA\u002Fexamples\u002F.turix_tmp\u002F\n\n# 4. Tail progress in real time\ntail -f your_dir\u002FTuriX-CUA\u002F.turix_tmp\u002Flogging.log\n\n# 5. Check current step count\nls your_dir\u002FTuriX-CUA\u002Fexamples\u002F.turix_tmp\u002Fbrain_llm_interactions.log_brain_*.txt | wc -l\n```\n",{"name":5,"description":7},"main","e7e336ac628c3e3bd771aff1aaa469c93778f94a","skills\u002Ftongyu-yan\u002Fturix-cua",0,"https:\u002F\u002Fgithub.com\u002Fkbarbel640-del\u002Fskills","https:\u002F\u002Fgithub.com\u002Fkbarbel640-del","2026-08-23 22:39:37",{"source":8,"skillName":5,"commands":52,"prompt":62,"zipAvailable":63},[53,56,59],{"tool":54,"command":55},"npx","npx skills add kbarbel640-del\u002Fskills --skill turix-mac",{"tool":57,"command":58},"bunx","bunx skills add kbarbel640-del\u002Fskills --skill turix-mac",{"tool":60,"command":61},"pnpm","pnpm dlx skills add kbarbel640-del\u002Fskills --skill turix-mac","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fkbarbel640-del\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Ftongyu-yan\u002Fturix-cua\u002FSKILL.md",true,[65,71,75],{"path":66,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":69,"sourceHtmlUrl":70},"_meta.json","REFERENCE","参考文档",989,"https:\u002F\u002Fgithub.com\u002Fkbarbel640-del\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Ftongyu-yan\u002Fturix-cua\u002F_meta.json",{"path":72,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":73,"sourceHtmlUrl":74},"README.md",2462,"https:\u002F\u002Fgithub.com\u002Fkbarbel640-del\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Ftongyu-yan\u002Fturix-cua\u002FREADME.md",{"path":76,"resourceType":77,"resourceTypeDesc":78,"sizeBytes":79,"sourceHtmlUrl":80},"scripts\u002Frun_turix.sh","SCRIPT","脚本",5427,"https:\u002F\u002Fgithub.com\u002Fkbarbel640-del\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Ftongyu-yan\u002Fturix-cua\u002Fscripts\u002Frun_turix.sh",[82,112,135,156,162,182],{"skillNo":83,"name":84,"nameZh":6,"description":85,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":86,"categories":87,"keywords":93,"publishedAt":39},"01M6G8J55967D6281ACMKAKTEQ","crowd-prompting","A marketplace where AI agents improve prompts, system instructions, tool descriptions, and other text-based content with domain expertise from real-world operations — and earn tokens for valuable contributions.","https:\u002F\u002Fgithub.com\u002Fkbarbel640-del\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fzzadrian\u002Fcrowd-prompting\u002FSKILL.md",[88],{"categoryNo":89,"categoryName":90,"taxonomyType":91,"taxonomyTypeDesc":92},"01JBSEEDCAT000000000000036","内容创作者","PROFESSION","适用职业",[94,95,31,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111],"crowd prompting","prompting","with","tokens","domain","world","prompts","operations","instructions","text","tool","system","improve","productivity","marketplace","https","content","ai agents",{"skillNo":113,"name":114,"nameZh":6,"description":115,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":116,"categories":117,"keywords":119,"publishedAt":39},"01M6G8J0PKPAYVET6BWE8W4C26","dilemm-ai","Compete in DilemmAI, the prisoner's dilemma AI arena at dilemm.ai. Use when an OpenClaw agent wants to create an account, design and submit strategy prompts for their bot, enter matchmaking, analyze historical matches and leaderboards, and iterate on bot strategy. Requires headless browser automation (no API available yet).","https:\u002F\u002Fgithub.com\u002Fkbarbel640-del\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fzyori\u002Fdilemm-ai\u002FSKILL.md",[118],{"categoryNo":18,"categoryName":19,"taxonomyType":20,"taxonomyTypeDesc":21},[120,31,121,122,123,124,125,27,126,127,128,129,130,131,132,133,134,100,33],"dilemm ai","arena","analyze","browser","openclaw agent","create","browser automation","bot","openclaw","design","strategy","headless","compete","submit","matchmaking",{"skillNo":136,"name":137,"nameZh":6,"description":138,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":139,"categories":140,"keywords":144,"publishedAt":39},"01M6G8HWAF3K9SW7NP3Z83CPJ6","arxiv-paper-reviews","Interact with arXiv Crawler API to fetch papers, read reviews, and submit comments. Use when working with arXiv papers, fetching paper lists by date\u002Fcategory\u002Finterest, viewing paper details with comments, or submitting paper reviews via API at http:\u002F\u002F150.158.152.82:8000.","https:\u002F\u002Fgithub.com\u002Fkbarbel640-del\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fzxrys\u002Fweak-accept\u002FSKILL.md",[141],{"categoryNo":142,"categoryName":143,"taxonomyType":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000021","资料检索与研究",[145,146,147,148,149,150,151,96,152,133,33,153,31,154,155],"arxiv paper reviews","reviews","arxiv","paper","crawler","papers","comments","fetch","read","http","via",{"skillNo":157,"name":137,"nameZh":6,"description":138,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":158,"categories":159,"keywords":161,"publishedAt":39},"01M6G8HQJ7N7WMY6YV7D7SSJ16","https:\u002F\u002Fgithub.com\u002Fkbarbel640-del\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fzxrys\u002Farxiv-paper-reviews\u002FSKILL.md",[160],{"categoryNo":142,"categoryName":143,"taxonomyType":20,"taxonomyTypeDesc":21},[148,147,146,145,155,150,151,96,149,133,33,152,153,31,154],{"skillNo":163,"name":164,"nameZh":6,"description":165,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":166,"categories":167,"keywords":168,"publishedAt":39},"01M6G8HM95Y64JNVFC2HN4E2BG","git-summary","Get a quick summary of the current Git repository including status, recent commits, branches, and contributors.","https:\u002F\u002Fgithub.com\u002Fkbarbel640-del\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fzweack\u002Fgit-summary\u002FSKILL.md",[],[169,170,171,172,173,174,175,176,177,178,179,180,181],"git","summary","git summary","commits","repository","win32","status","linux","contributors","true","quick","get","branches",{"skillNo":183,"name":184,"nameZh":6,"description":185,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":186,"categories":187,"keywords":192,"publishedAt":39},"01M6G8HGEGF05CNEJCGVYFSC4R","agent-constitution","Interact with AgentConstitution governance contracts on Base Sepolia. Check compliance, read rules, log actions, query governance state.","https:\u002F\u002Fgithub.com\u002Fkbarbel640-del\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fztsalexey\u002Fagent-constitution\u002FSKILL.md",[188,189],{"categoryNo":18,"categoryName":19,"taxonomyType":20,"taxonomyTypeDesc":21},{"categoryNo":190,"categoryName":191,"taxonomyType":91,"taxonomyTypeDesc":92},"01JBSEEDCAT000000000000037","律师与法务",[193,194,195,196,96,197,198,44,199,153,200,201,202,203,204,109,205,206,207],"agent constitution","constitution","contracts","tree","query","base","log","check","Rules","projects","compliance","actions","state","governance","github"]