From 96142c3a40437cecef8efb93448a8f224df4f334 Mon Sep 17 00:00:00 2001 From: Masha_Rudenko Date: Mon, 13 Jul 2026 14:57:38 +0300 Subject: [PATCH 1/4] [add] MCP server guide for To Do List docs - new guide covering task hierarchy, projects, selection, and REST backend scenarios for the DHTMLX MCP server - setup instructions for Claude Code, Cursor, Antigravity, and ChatGPT - registered guides/mcp-server in the Guides sidebar category --- docs/guides/mcp-server.md | 239 ++++++++++++++++++++++++++++++++++++++ sidebars.js | 1 + 2 files changed, 240 insertions(+) create mode 100644 docs/guides/mcp-server.md diff --git a/docs/guides/mcp-server.md b/docs/guides/mcp-server.md new file mode 100644 index 0000000..ea7dd52 --- /dev/null +++ b/docs/guides/mcp-server.md @@ -0,0 +1,239 @@ +--- +sidebar_label: DHTMLX MCP server +title: Connect AI tools to DHTMLX To Do List with MCP +description: Connect AI coding assistants to live DHTMLX To Do List documentation via the MCP server. Covers tasks, subtasks, projects, priorities, and the REST backend. +--- + +# Connect AI tools to DHTMLX To Do List with MCP + +DHTMLX To Do List looks like a simple checklist on the surface, but it packs in [nested subtasks](guides/task_operations.md#adding-a-new-task), [priority hotkeys](/#prioritizing-a-task), [drag-and-drop reordering](guides/configuration.md#drag-n-drop), [hashtag-based filtering](guides/inline_editing.md#hashtags), and a REST backend that chains through the [event bus](api/internal/setnext_method.md). When an AI assistant answers from stale training data instead of querying the DHTMLX MCP server, it tends to invent method signatures such as [`copyTask()`](api/methods/copytask_method.md) or [`indentTask()`](api/methods/indenttask_method.md), misplace event payloads, or suggest [`taskShape`](api/configs/taskshape_config.md) options that no longer exist. + +That's where the DHTMLX MCP server helps: it gives the assistant direct access to the live To Do List documentation before it answers. Point it at [task operations](guides/task_operations.md), [multiselection and bulk actions](guides/multiselection.md), the [REST backend integration](guides/working_with_server.md), or [task and project configuration](guides/configuration.md), and the assistant retrieves the current API surface instead of guessing. + +#### MCP endpoint + +~~~jsx +https://docs.dhtmlx.com/mcp +~~~ + +:::note +The DHTMLX MCP server covers all major DHTMLX products, not only DHTMLX To Do List. The same endpoint and configuration instructions work regardless of which DHTMLX component you are building with. +::: + +## Where MCP server helps with To Do List + +The MCP server indexes the full DHTMLX To Do List documentation. Common scenarios include: + +- Looking up the current API for [To Do List methods](api/overview/methods_overview.md), [events](api/overview/events_overview.md), [properties](api/overview/configs_overview.md), and the matching [Toolbar API](category/toolbar-properties.md). +- Generating ready-to-run [initialization](guides/initialization.md) and [configuration](guides/configuration.md) code for a specific set of tasks, projects, and users. +- Managing task hierarchy and bulk actions, such as [adding, moving, and deleting tasks](guides/task_operations.md), changing the [indent level](guides/task_operations.md#changing-the-indent-level-of-a-task), and running operations over [multiple selected tasks](guides/multiselection.md). +- Working with [projects](guides/project_operations.md) and the [project object](guides/project_object_operations.md), including switching the active project and moving tasks between projects. +- Assigning [task performers](guides/task_users.md), setting due dates, and configuring [priorities](api/configs/priorities_config.md). +- Connecting `RestDataProvider` to a [REST backend](guides/working_with_server.md), including the multiuser mode and the event bus order set through `api.setNext()`. +- Configuring [sorting and filtering](guides/sorting_filtering_tasks.md), [hiding completed tasks](guides/hide_completed_tasks.md), and [read-only mode](guides/readonly_mode.md). +- Handling [inline editing](guides/inline_editing.md) and [keyboard shortcuts](guides/keyboard_navigation.md) for tasks and projects. +- Exploring [localization](guides/localization.md), [stylization](guides/stylization.md), and integration with [React](guides/integration_with_react.md), [Vue](guides/integration_with_vue.md), [Angular](guides/integration_with_angular.md), and [Svelte](guides/integration_with_svelte.md). + +## How DHTMLX MCP server works + +The DHTMLX MCP server combines Retrieval-Augmented Generation (RAG) with the Model Context Protocol (MCP) so that AI assistants can query documentation on demand rather than relying solely on training data. + +For example, when you ask *"How do I chain RestDataProvider into the event bus with api.setNext() so add-task and move-task operations reach my server?"*, the assistant sends the prompt via the MCP endpoint. The server matches it against the working-with-server documentation, retrieves the relevant reference pages, and returns them as context. The assistant then generates code based on the current API rather than a training snapshot. + +## Connect the MCP server to your AI tool + +Whether you are scaffolding a new To Do List project or wiring `RestDataProvider` into an existing backend, most AI development tools let you add MCP endpoints through a CLI command or a JSON configuration file. In either case, you register the server URL. + +~~~jsx +https://docs.dhtmlx.com/mcp +~~~ + +Below are setup instructions for commonly used tools. + +### Claude Code + +:::info +The [official documentation](https://code.claude.com/docs/en/mcp) covers all options for connecting Claude Code with MCP servers. +::: + +To register the server from the command line, run: + +~~~jsx +claude mcp add --transport http dhtmlx-mcp https://docs.dhtmlx.com/mcp +~~~ + +For manual setup, add the following to your `mcp.json`: + +~~~jsx +{ + "mcpServers": { + "dhtmlx-mcp": { + "type": "http", + "url": "https://docs.dhtmlx.com/mcp" + } + } +} +~~~ + +### Cursor + +:::info +The [official documentation](https://cursor.com/en-US/docs/mcp) covers all MCP configuration options for Cursor. +::: + +Steps to add the server: + +1. Open Settings (`Cmd+Shift+J` on Mac, `Ctrl+Shift+J` on Windows/Linux) +2. Go to **Tools & MCP** +3. Click **Add Custom MCP** +4. Paste the following config: + +~~~jsx +{ + "mcpServers": { + "dhtmlx-mcp": { + "url": "https://docs.dhtmlx.com/mcp" + } + } +} +~~~ + +### Google Antigravity + +#### Antigravity 2.0 + +:::info +Refer to the [official documentation](https://antigravity.google/docs/mcp) for full details on MCP server integration in Antigravity. +::: + +These are the steps to complete for connecting DHTMLX MCP server with Google Antigravity: + +1. Open the command palette +2. Type "mcp add" +3. Select "HTTP" +4. Provide the following values: +- Name: +~~~jsx +dhtmlx-mcp +~~~ +- URL: +~~~jsx +https://docs.dhtmlx.com/mcp +~~~ + +#### Antigravity CLI + +:::info +Check the [related guide](https://antigravity.google/docs/gcli-migration#mcp-config-formatting-changes) to learn about migration from Gemini CLI to Antigravity CLI. +::: + +To connect the DHTMLX MCP server to Antigravity CLI, create `mcp_config.json` in one of these locations: + +- Global: `~/.gemini/config/mcp_config.json` +- Workspace: `.agents/mcp_config.json` + +Add the following configuration: + +~~~jsx +{ + "mcpServers": { + "dhtmlx-mcp": { + "serverUrl": "https://docs.dhtmlx.com/mcp" + } + } +} +~~~ + +Then run `agy` in the terminal. + +### ChatGPT + +:::info +The [official documentation](https://developers.openai.com/api/docs/guides/tools-connectors-mcp) covers MCP connector setup for ChatGPT. +::: + +Steps to configure the connector: + +1. Go to **Settings** → **Apps & Connectors** +2. Click **Advanced settings** +3. Enable **Developer mode** +4. Return to **Apps & Connectors** and click "Create" +5. Fill in the connector details: +- Name: +~~~jsx +dhtmlx-mcp +~~~ +- URL: +~~~jsx +https://docs.dhtmlx.com/mcp +~~~ +- Authentication: `No authentication` +6. Click **Create** + +After you create the connector, ChatGPT pulls documentation from the MCP server during conversations. + +:::warning +For intensive coding workflows, other MCP-aware tools may be a better fit. +::: + +### Other tools + +Many modern AI coding tools expose MCP support under labels such as "Model Context Protocol", "Context Sources", or similar. Add `https://docs.dhtmlx.com/mcp` as a custom source in the relevant settings panel. + +## Privacy and data handling + +The MCP server is a hosted service: nothing runs on your machine, no files from your environment are read, and no personal user data is stored. + +For debugging and service improvement, queries sent to the server may be logged. + +If your organization needs stricter guarantees, ask about a commercial deployment with query logging turned off by writing to `info@dhtmlx.com`. + +## Example prompts for To Do List with AI + +Once you connect the MCP server, phrase your prompts around a concrete goal so the assistant knows which part of the To Do List API to retrieve. The prompts below are organized by task type. Copy and adapt them as needed. + +**Tasks and hierarchy** + +~~~ +How do I add a subtask under a specific task using addTask() in DHTMLX To Do List? +~~~ +~~~ +How do I indent a task with indentTask() to make it a subtask of the task above it? Use the docs. +~~~ +~~~ +How do I copy a task with all its subtasks and paste it into another project with copyTask() and pasteTask()? +~~~ + +**Projects** + +~~~ +How do I switch the active project with setProject() and move a task into it? +~~~ +~~~ +What happens to a project's tasks when I delete it with deleteProject() in DHTMLX To Do List? +~~~ + +**Selection and bulk operations** + +~~~ +How do I select several tasks and mark them all as complete at once in DHTMLX To Do List? +~~~ +~~~ +How do I use eachSelected() to delete all currently selected tasks? Use the docs. +~~~ + +**Server integration** + +~~~ +How do I connect RestDataProvider to my Node.js backend and load tasks on initialization? +~~~ +~~~ +How do I set up the multiuser mode with RemoteEvents so task changes sync across clients in real time? +~~~ + +## Tips for effective To Do List prompts + +- **Name the exact method.** DHTMLX To Do List exposes several similarly named method pairs (`checkTask()`/`uncheckTask()`, `indentTask()`/`unindentTask()`, `hideCompletedTasks()`/`showCompletedTasks()`). State the method name so the assistant retrieves the right reference instead of guessing the parameters. +- **Say whether the target is a task, a project, or a user.** Many methods take a similar id-based object (for example, `getTask()` vs. `getProject()`). Naming the target narrows retrieval to the correct part of the API. +- **Describe the data shape.** Prompts like "a task with subtasks and a due date" or "a task assigned to several users" retrieve more precise documentation than a generic "a task". This matters most when preparing data to load or configuring `taskShape`. +- **Add "Use the docs"** to your prompt. This phrase signals that the assistant should trigger an MCP lookup instead of answering from training data alone. It helps most with event payloads (for example, `open-menu` or `edit-item`) and REST backend wiring, both of which change between versions. diff --git a/sidebars.js b/sidebars.js index d5843c2..1ef1d7e 100644 --- a/sidebars.js +++ b/sidebars.js @@ -490,6 +490,7 @@ module.exports = { "guides/customization", "guides/stylization", //"guides/integration", + "guides/mcp-server", ] } ] From aa8d51ba0cd70cf8b6b6c4135097bd8ae6c6bb72 Mon Sep 17 00:00:00 2001 From: Masha_Rudenko Date: Mon, 20 Jul 2026 13:29:44 +0300 Subject: [PATCH 2/4] [update] rework To Do List MCP guide for uniqueness and SEO - reworded title, description, and all six section headings that were duplicated (or near-duplicated) across other DHTMLX product MCP guides - reworded the intro's AI-failure-mode sentence and MCP-server-as- solution clause, and the "Where MCP helps"/"How it works" opening sentences, for the same reason - added a Search vs Inference workflow explanation to the "How it works" section - updated the ChatGPT setup section with the current OpenAI docs URL - changed the MCP endpoint block from a heading to bold text to avoid skipping a heading level (H1 straight to H4) - added inbound links to the guide from how_to_start.md and overview.md --- docs/guides/mcp-server.md | 36 ++++++++++++++++++------------------ docs/how_to_start.md | 3 ++- docs/overview.md | 2 +- 3 files changed, 21 insertions(+), 20 deletions(-) diff --git a/docs/guides/mcp-server.md b/docs/guides/mcp-server.md index ea7dd52..68f087d 100644 --- a/docs/guides/mcp-server.md +++ b/docs/guides/mcp-server.md @@ -1,16 +1,16 @@ --- sidebar_label: DHTMLX MCP server -title: Connect AI tools to DHTMLX To Do List with MCP -description: Connect AI coding assistants to live DHTMLX To Do List documentation via the MCP server. Covers tasks, subtasks, projects, priorities, and the REST backend. +title: DHTMLX To Do List MCP server keeps AI answers accurate +description: Tasks, subtasks, projects, priorities, and REST sync in DHTMLX To Do List all stay current for AI assistants through the MCP server. --- -# Connect AI tools to DHTMLX To Do List with MCP +# DHTMLX To Do List MCP server: docs that stay current for AI -DHTMLX To Do List looks like a simple checklist on the surface, but it packs in [nested subtasks](guides/task_operations.md#adding-a-new-task), [priority hotkeys](/#prioritizing-a-task), [drag-and-drop reordering](guides/configuration.md#drag-n-drop), [hashtag-based filtering](guides/inline_editing.md#hashtags), and a REST backend that chains through the [event bus](api/internal/setnext_method.md). When an AI assistant answers from stale training data instead of querying the DHTMLX MCP server, it tends to invent method signatures such as [`copyTask()`](api/methods/copytask_method.md) or [`indentTask()`](api/methods/indenttask_method.md), misplace event payloads, or suggest [`taskShape`](api/configs/taskshape_config.md) options that no longer exist. +DHTMLX To Do List looks like a simple checklist on the surface, but it packs in [nested subtasks](guides/task_operations.md#adding-a-new-task), [priority hotkeys](/#prioritizing-a-task), [drag-and-drop reordering](guides/configuration.md#drag-n-drop), [hashtag-based filtering](guides/inline_editing.md#hashtags), and a REST backend that chains through the [event bus](api/internal/setnext_method.md). Skip the MCP server and an assistant falls back on stale training data: it invents method signatures for [`copyTask()`](api/methods/copytask_method.md) or [`indentTask()`](api/methods/indenttask_method.md) that don't match the real API, misplaces event payloads, or reaches for [`taskShape`](api/configs/taskshape_config.md) options that no longer exist. -That's where the DHTMLX MCP server helps: it gives the assistant direct access to the live To Do List documentation before it answers. Point it at [task operations](guides/task_operations.md), [multiselection and bulk actions](guides/multiselection.md), the [REST backend integration](guides/working_with_server.md), or [task and project configuration](guides/configuration.md), and the assistant retrieves the current API surface instead of guessing. +That's where the DHTMLX MCP server helps: it lets the assistant check the live To Do List documentation before it answers. Point it at [task operations](guides/task_operations.md), [multiselection and bulk actions](guides/multiselection.md), the [REST backend integration](guides/working_with_server.md), or [task and project configuration](guides/configuration.md), and the assistant retrieves the current API surface instead of guessing. -#### MCP endpoint +**MCP endpoint** ~~~jsx https://docs.dhtmlx.com/mcp @@ -20,9 +20,9 @@ https://docs.dhtmlx.com/mcp The DHTMLX MCP server covers all major DHTMLX products, not only DHTMLX To Do List. The same endpoint and configuration instructions work regardless of which DHTMLX component you are building with. ::: -## Where MCP server helps with To Do List +## MCP server support across the To Do List API -The MCP server indexes the full DHTMLX To Do List documentation. Common scenarios include: +The server's index covers the full DHTMLX To Do List documentation, including: - Looking up the current API for [To Do List methods](api/overview/methods_overview.md), [events](api/overview/events_overview.md), [properties](api/overview/configs_overview.md), and the matching [Toolbar API](category/toolbar-properties.md). - Generating ready-to-run [initialization](guides/initialization.md) and [configuration](guides/configuration.md) code for a specific set of tasks, projects, and users. @@ -34,13 +34,13 @@ The MCP server indexes the full DHTMLX To Do List documentation. Common scenario - Handling [inline editing](guides/inline_editing.md) and [keyboard shortcuts](guides/keyboard_navigation.md) for tasks and projects. - Exploring [localization](guides/localization.md), [stylization](guides/stylization.md), and integration with [React](guides/integration_with_react.md), [Vue](guides/integration_with_vue.md), [Angular](guides/integration_with_angular.md), and [Svelte](guides/integration_with_svelte.md). -## How DHTMLX MCP server works +## Two ways the MCP server can respond -The DHTMLX MCP server combines Retrieval-Augmented Generation (RAG) with the Model Context Protocol (MCP) so that AI assistants can query documentation on demand rather than relying solely on training data. +A Retrieval-Augmented Generation (RAG) index backs the Model Context Protocol (MCP) endpoint, letting an assistant check current To Do List documentation instead of relying only on what it learned during training. The index serves two workflows: *Search*, which hands back matching reference pages for the assistant to build an answer from, and *Inference*, which reads those pages itself and delivers the finished answer. -For example, when you ask *"How do I chain RestDataProvider into the event bus with api.setNext() so add-task and move-task operations reach my server?"*, the assistant sends the prompt via the MCP endpoint. The server matches it against the working-with-server documentation, retrieves the relevant reference pages, and returns them as context. The assistant then generates code based on the current API rather than a training snapshot. +For example, when you ask *"How do I chain RestDataProvider into the event bus with api.setNext() so add-task and move-task operations reach my server?"*, the assistant sends the prompt via the MCP endpoint. *Search* matches it against the working-with-server documentation, retrieves the relevant reference pages, and returns them as context; the assistant then generates code based on the current API rather than a training snapshot. *Inference* suits a narrower question with one correct answer, such as confirming which event fires first in the chain, and returns that answer directly instead of the source pages. -## Connect the MCP server to your AI tool +## Setting up the MCP connection Whether you are scaffolding a new To Do List project or wiring `RestDataProvider` into an existing backend, most AI development tools let you add MCP endpoints through a CLI command or a JSON configuration file. In either case, you register the server URL. @@ -149,7 +149,7 @@ Then run `agy` in the terminal. ### ChatGPT :::info -The [official documentation](https://developers.openai.com/api/docs/guides/tools-connectors-mcp) covers MCP connector setup for ChatGPT. +The [official documentation](https://help.openai.com/en/articles/12584461-developer-mode-and-mcp-apps-in-chatgpt) covers MCP connector setup for ChatGPT. ::: Steps to configure the connector: @@ -172,7 +172,7 @@ https://docs.dhtmlx.com/mcp After you create the connector, ChatGPT pulls documentation from the MCP server during conversations. -:::warning +:::info For intensive coding workflows, other MCP-aware tools may be a better fit. ::: @@ -180,7 +180,7 @@ For intensive coding workflows, other MCP-aware tools may be a better fit. Many modern AI coding tools expose MCP support under labels such as "Model Context Protocol", "Context Sources", or similar. Add `https://docs.dhtmlx.com/mcp` as a custom source in the relevant settings panel. -## Privacy and data handling +## How the MCP server treats your data The MCP server is a hosted service: nothing runs on your machine, no files from your environment are read, and no personal user data is stored. @@ -188,9 +188,9 @@ For debugging and service improvement, queries sent to the server may be logged. If your organization needs stricter guarantees, ask about a commercial deployment with query logging turned off by writing to `info@dhtmlx.com`. -## Example prompts for To Do List with AI +## To Do List prompts to copy and adapt -Once you connect the MCP server, phrase your prompts around a concrete goal so the assistant knows which part of the To Do List API to retrieve. The prompts below are organized by task type. Copy and adapt them as needed. +State the concrete To Do List goal in your prompt and the assistant identifies the right slice of the API. The groups below sort the examples by task; adapt any of them to your own data. **Tasks and hierarchy** @@ -231,7 +231,7 @@ How do I connect RestDataProvider to my Node.js backend and load tasks on initia How do I set up the multiuser mode with RemoteEvents so task changes sync across clients in real time? ~~~ -## Tips for effective To Do List prompts +## Making To Do List prompts more precise - **Name the exact method.** DHTMLX To Do List exposes several similarly named method pairs (`checkTask()`/`uncheckTask()`, `indentTask()`/`unindentTask()`, `hideCompletedTasks()`/`showCompletedTasks()`). State the method name so the assistant retrieves the right reference instead of guessing the parameters. - **Say whether the target is a task, a project, or a user.** Many methods take a similar id-based object (for example, `getTask()` vs. `getProject()`). Naming the target narrows retrieval to the correct part of the API. diff --git a/docs/how_to_start.md b/docs/how_to_start.md index 416da26..71123f6 100644 --- a/docs/how_to_start.md +++ b/docs/how_to_start.md @@ -250,4 +250,5 @@ If you feel like diving deeper into the world of To Do List, these are suggestio - [To Do List overview](/) - [Guides](category/guides.md) -- [API](category/api.md) \ No newline at end of file +- [API](category/api.md) +- [DHTMLX MCP server](guides/mcp-server.md) \ No newline at end of file diff --git a/docs/overview.md b/docs/overview.md index 49eb2e5..146c840 100644 --- a/docs/overview.md +++ b/docs/overview.md @@ -141,4 +141,4 @@ Read the [**keyboard shortcuts**](api/events/keypressontodo_event.md#keyboard-sh ## What's next -Once you have a short overview of To Do List you are ready to learn how to display the component on the page. Follow the directions given in the [How to start](how_to_start.md) article. \ No newline at end of file +Once you have a short overview of To Do List you are ready to learn how to display the component on the page. Follow the directions given in the [How to start](how_to_start.md) article. If you build with an AI coding assistant, connect it to the live documentation through the [DHTMLX MCP server](guides/mcp-server.md). \ No newline at end of file From 8f7bef1175ba641d69e56a76d998297a209c9f1f Mon Sep 17 00:00:00 2001 From: Masha_Rudenko Date: Mon, 20 Jul 2026 17:25:35 +0300 Subject: [PATCH 3/4] [dev] clarify mcp-server guide wording - reworded RAG/MCP explanation for clarity - adjusted prompt-writing guidance wording --- docs/guides/mcp-server.md | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/docs/guides/mcp-server.md b/docs/guides/mcp-server.md index 68f087d..7c46363 100644 --- a/docs/guides/mcp-server.md +++ b/docs/guides/mcp-server.md @@ -36,9 +36,9 @@ The server's index covers the full DHTMLX To Do List documentation, including: ## Two ways the MCP server can respond -A Retrieval-Augmented Generation (RAG) index backs the Model Context Protocol (MCP) endpoint, letting an assistant check current To Do List documentation instead of relying only on what it learned during training. The index serves two workflows: *Search*, which hands back matching reference pages for the assistant to build an answer from, and *Inference*, which reads those pages itself and delivers the finished answer. +The assistant chooses between two workflows depending on what's being asked. *Search* returns the matching To Do List reference pages so the assistant can compose the answer; *Inference* reads those pages itself and returns the finished answer. Both run on a Retrieval-Augmented Generation (RAG) index built from the documentation, reached through the Model Context Protocol (MCP). -For example, when you ask *"How do I chain RestDataProvider into the event bus with api.setNext() so add-task and move-task operations reach my server?"*, the assistant sends the prompt via the MCP endpoint. *Search* matches it against the working-with-server documentation, retrieves the relevant reference pages, and returns them as context; the assistant then generates code based on the current API rather than a training snapshot. *Inference* suits a narrower question with one correct answer, such as confirming which event fires first in the chain, and returns that answer directly instead of the source pages. +For example, when you ask *"How do I chain RestDataProvider into the event bus with api.setNext() so add-task and move-task operations reach my server?"*, the assistant sends the prompt via the MCP endpoint. *Search* matches it against the working-with-server documentation, retrieves the relevant reference pages, and returns them as context; the assistant then produces code that matches the API's current state instead of an old training snapshot. *Inference* suits a narrower question with one correct answer, such as confirming which event fires first in the chain, and returns that answer directly instead of the source pages. ## Setting up the MCP connection @@ -190,7 +190,7 @@ If your organization needs stricter guarantees, ask about a commercial deploymen ## To Do List prompts to copy and adapt -State the concrete To Do List goal in your prompt and the assistant identifies the right slice of the API. The groups below sort the examples by task; adapt any of them to your own data. +These prompts work best when they name a To Do List object directly, a task, a project, a user, rather than describing it in the abstract. Each group below targets one kind of operation; adapt the wording to your own data. **Tasks and hierarchy** From c65462032afc48cc19ad8eee9038b1f25e97cc43 Mon Sep 17 00:00:00 2001 From: Masha_Rudenko Date: Mon, 27 Jul 2026 17:57:20 +0300 Subject: [PATCH 4/4] [update] vary Setup-section wording in MCP guide - reworded the Setup section's intro sentence, five tool-specific info boxes (Claude Code, Cursor, Antigravity 2.0, Antigravity CLI, ChatGPT), and the "Other tools" paragraph so they no longer read as verbatim-identical to the same content in the Kanban, Spreadsheet, Diagram, Booking, and Pivot MCP guides - same links and facts, different sentence construction throughout --- docs/guides/mcp-server.md | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-) diff --git a/docs/guides/mcp-server.md b/docs/guides/mcp-server.md index 7c46363..0690c15 100644 --- a/docs/guides/mcp-server.md +++ b/docs/guides/mcp-server.md @@ -42,7 +42,7 @@ For example, when you ask *"How do I chain RestDataProvider into the event bus w ## Setting up the MCP connection -Whether you are scaffolding a new To Do List project or wiring `RestDataProvider` into an existing backend, most AI development tools let you add MCP endpoints through a CLI command or a JSON configuration file. In either case, you register the server URL. +Setting up a fresh To Do List project or hooking into an existing `RestDataProvider` backend both start the same way: register the MCP endpoint with your tool, via a CLI command or a JSON configuration entry, depending on what that tool expects. ~~~jsx https://docs.dhtmlx.com/mcp @@ -53,7 +53,7 @@ Below are setup instructions for commonly used tools. ### Claude Code :::info -The [official documentation](https://code.claude.com/docs/en/mcp) covers all options for connecting Claude Code with MCP servers. +For the complete rundown of Claude Code's MCP setup, check the [official documentation](https://code.claude.com/docs/en/mcp). ::: To register the server from the command line, run: @@ -78,7 +78,7 @@ For manual setup, add the following to your `mcp.json`: ### Cursor :::info -The [official documentation](https://cursor.com/en-US/docs/mcp) covers all MCP configuration options for Cursor. +Step-by-step MCP configuration for Cursor lives in the [official documentation](https://cursor.com/en-US/docs/mcp). ::: Steps to add the server: @@ -103,7 +103,7 @@ Steps to add the server: #### Antigravity 2.0 :::info -Refer to the [official documentation](https://antigravity.google/docs/mcp) for full details on MCP server integration in Antigravity. +Antigravity documents its MCP server integration in the [official documentation](https://antigravity.google/docs/mcp). ::: These are the steps to complete for connecting DHTMLX MCP server with Google Antigravity: @@ -124,7 +124,7 @@ https://docs.dhtmlx.com/mcp #### Antigravity CLI :::info -Check the [related guide](https://antigravity.google/docs/gcli-migration#mcp-config-formatting-changes) to learn about migration from Gemini CLI to Antigravity CLI. +For Gemini CLI to Antigravity CLI migration steps, check the [related guide](https://antigravity.google/docs/gcli-migration#mcp-config-formatting-changes). ::: To connect the DHTMLX MCP server to Antigravity CLI, create `mcp_config.json` in one of these locations: @@ -149,7 +149,7 @@ Then run `agy` in the terminal. ### ChatGPT :::info -The [official documentation](https://help.openai.com/en/articles/12584461-developer-mode-and-mcp-apps-in-chatgpt) covers MCP connector setup for ChatGPT. +ChatGPT's own documentation lays out the complete [MCP connector setup process](https://help.openai.com/en/articles/12584461-developer-mode-and-mcp-apps-in-chatgpt). ::: Steps to configure the connector: @@ -178,7 +178,7 @@ For intensive coding workflows, other MCP-aware tools may be a better fit. ### Other tools -Many modern AI coding tools expose MCP support under labels such as "Model Context Protocol", "Context Sources", or similar. Add `https://docs.dhtmlx.com/mcp` as a custom source in the relevant settings panel. +Most other AI coding tools support MCP too, typically under a settings label like "Model Context Protocol" or "Context Sources." Add `https://docs.dhtmlx.com/mcp` there as a custom source. ## How the MCP server treats your data