System Setup & Configuration
Complete first-launch setup, understand the main interface, and prepare to use private libraries and shared spaces.
1.1 First Launch
Launch Process
When launching MeshWork for the first time, the application will guide you through the following steps:
- Select or create a private library: Choose an existing directory as your private library, or create a new private library directory. The private library is recommended for storing your long-term notes, drafts, materials, and unpublished ideas.
- Enter the main workspace: By default, the private library opens first, as it's the most stable personal entry point. You can see the private library's document tree in the upper half of the sidebar.
- Add shared spaces later: When team collaboration is needed, add shared spaces through the
+button at the bottom of the sidebar. Git configuration can be set up later for shared spaces.
You don't need to configure a remote repository from the start. MeshWork's design principle is to let you start writing and organizing first, then gradually enable sharing capabilities.
Main Interface Structure
After entering the main workspace, you'll see the following interface structure:
- Sidebar: The left side is divided into upper and lower parts. The upper part shows the private library and its document tree. The lower part shows the list of shared spaces. Below the shared space area is a
+button for adding new shared spaces. - Main Content Area: The middle area displays the content of the currently selected document or discussion area.
- Right Sidebar: When editing documents, the right sidebar can open the AI assistant to provide intelligent assistance.
1.2 Workspace Management
Understanding Workspace Structure
The sidebar of the main workspace interface displays all added content. The sidebar is divided into two areas:
- Upper part: Fixed display of the private library and its document tree.
- Lower part: Display of all added shared spaces.
The workspace management page is used to add physical directories (private libraries or shared spaces). Once added, these contents appear directly in the main interface's left sidebar without requiring additional switching operations.
Adding Shared Spaces
Add shared spaces through the + button at the bottom of the sidebar:
- Click the
+button at the bottom of the sidebar. - Select "Create Shared Space" or "Import Shared Space".
- When creating, specify the directory location and name. You can choose to use a template.
- When importing, specify the existing shared space directory.
- After adding, the shared space will immediately appear in the shared space list in the sidebar.
When creating a shared space, Git configuration is optional. You can use it locally first, then configure Git for team collaboration when needed.
Removing Shared Spaces
To remove a shared space from the sidebar:
- Right-click on the shared space you want to remove in the sidebar.
- Select "Remove from Workspace" option.
- After confirming removal, the shared space will no longer appear in the workspace sidebar.
Removing a shared space doesn't delete data on disk, it only hides it from the workspace view. You can re-add the shared space at any time.
1.3 Application Settings
MeshWork's application settings page is divided into multiple parts. You can open the settings window from the menu bar. The left navigation bar is arranged in the following order: Appearance, Document, Discussion, Network Proxy, AI, Git, About.
1.3.1 Appearance Settings
Customize interface display in "Settings → Appearance":
- Theme: Choose to follow system, light, or dark theme. Following the system theme automatically switches based on macOS appearance settings.
- Language: Choose interface language (Simplified Chinese or English). AI agents will automatically adjust response language based on the current application language.
- Interface Density: Adjust interface compactness. Choose between comfortable, compact, or relaxed, affecting spacing and element sizes.
- Interface Font Size: Adjust display size of interface text. Choose between small, medium, or large.
1.3.2 Document Settings
Configure document assistant behavior in "Settings → Document". These configurations are application-global and will affect all document editing pages in both private and public libraries:
- Custom Prompts: Override default document assistant prompts to make AI provide help according to your preferences.
- Custom Commands: Add your frequently used shortcut commands, such as
/summarize(summarize current document),/action(extract action items), etc. When typing/, the available command list will be displayed.
Document configuration is application-wide configuration, not belonging to any private or public library, and not written to resource directories. These configurations affect AI assistant behavior for all documents.
1.3.3 Discussion Settings
Configure default behavior for discussion area topics in "Settings → Discussion":
- Auto Session Duration: Set the default auto session window duration for discussion area topics, default is 2 minutes, range is 2-10 minutes. During this time, your continued messages will automatically be sent to the last responding agent without needing to @ repeatedly.
- Prompt: Set the default prompt for new topics, defining the agent's role and behavior in discussions.
- Branch Summary Prompt: Customize automatic summary prompts for topic branches, used to generate brief descriptions of branches.
- Custom Commands: Add topic-level custom commands, these commands are only available in discussion areas.
- Tool Call Limit Per Prompt: Set the maximum number of tool calls an agent can make in a single response, preventing accidental excessive resource consumption.
Discussion area topics can override these default settings. In individual topic settings, you can specify different prompts, session durations, and commands for that topic. When left empty, the default values from here are inherited.
1.3.4 Network Proxy Settings
Configure proxy services in "Settings → Network Proxy". This is useful for users who need to access AI services or search services through proxies:
- Add Proxy Configuration: Click "Add" button, enter proxy name and address (such as
127.0.0.1:7890). - Proxy Type: Supports HTTP, HTTPS, and SOCKS5 proxies.
- Authentication: If the proxy requires authentication, enter username and password.
- Apply Proxy: In AI Provider or network search service configuration, choose which proxy configuration to use.
Proxy passwords are securely stored on your local device. Please ensure proxy servers are trustworthy to avoid leaking sensitive information through the proxy.
1.3.5 Git Settings
Understanding MeshWork's Git Model
Before configuring Git, you need to understand the relationship between MeshWork and Git:
- The workspace itself is not a Git repository; Git is only bound to real data boundaries.
- The private library can choose to enable or disable Git. Even when enabled, it only manages the private library's own content.
- Each shared space can have its own Git configuration; Git only acts on the current shared space.
- Resources of the same type can share the same remote repository, distinguished by different branches.
MeshWork treats Git as a hidden, reliable foundation. You don't need to understand concepts like commit, branch, or merge in daily use. MeshWork automatically handles version history, change tracking, collaborative sharing, and recovery capabilities.
1.3.6 AI Settings
Configure AI agent behavior in "Settings → AI". MeshWork supports multiple AI services and search services through compatible protocols. You can add and configure them as needed.
AI Protocols and Compatible Providers
MeshWork currently includes two AI protocol formats: OpenAI Compatible and Anthropic Compatible. In settings, "OpenAI" and "Anthropic" describe the request format and protocol type; they do not restrict you to those two companies. Any third-party service that exposes a compatible API can be configured with its API key, Base URL, and model name.
The following are common compatible provider examples. Model names, tool calling, streaming, thinking mode, and other capabilities vary by provider, so check the provider documentation when configuring them:
| Provider | Compatible Protocol | Docs / Platform | Common Base URL |
|---|---|---|---|
| OpenAI | OpenAI Compatible | platform.openai.com/docs | https://api.openai.com/v1 |
| Anthropic Claude | Anthropic Compatible | docs.anthropic.com | https://api.anthropic.com/v1 |
| OpenRouter | OpenAI Compatible | openrouter.ai/docs | https://openrouter.ai/api/v1 |
| DeepSeek | OpenAI Compatible / Anthropic Compatible | api-docs.deepseek.com | https://api.deepseek.com; Anthropic format uses https://api.deepseek.com/anthropic |
| Alibaba Cloud Model Studio / DashScope | OpenAI Compatible / Anthropic Compatible | help.aliyun.com/zh/model-studio | Choose by region, for example https://dashscope.aliyuncs.com/compatible-mode/v1 |
| Zhipu AI / BigModel | OpenAI Compatible | docs.bigmodel.cn | https://open.bigmodel.cn/api/paas/v4 |
| Moonshot AI / Kimi | OpenAI Compatible | platform.kimi.ai/docs | https://api.moonshot.ai/v1 |
| SiliconFlow | OpenAI Compatible | docs.siliconflow.cn | https://api.siliconflow.cn/v1 |
| Mistral AI | OpenAI Compatible | docs.mistral.ai | https://api.mistral.ai/v1 |
Configuration Options:
- Provider / Protocol: Choose OpenAI Compatible or Anthropic Compatible based on the target service's API format.
- API Key: Enter the API key obtained from the corresponding platform.
- Base URL: Enter the provider's compatible API root. Usually do not include
/chat/completionsor a model-specific path. - Model: Enter or choose the model name from the provider documentation. Default examples are
gpt-4oandclaude-3-5-sonnet-20241022; change this to the provider's model ID when using a third-party service. - Context Window Size: Set maximum context window (unit KB), default is 200KB, range is 32KB-2048KB.
- Proxy Settings: If proxy access is needed, select the corresponding proxy configuration.
- Set as Default: Mark as default provider, AI assistant will prioritize using it.