mor-blogger-theme-editor-mcp

# MorBlogger MCP Engine A headless, zero-dependency stdio bridge that connects LLM clients directly to the MorBlogger Rust compiler. Instead of an AI guessing how your theme architecture works, this Model Context Protocol (MCP) server allows your AI assistant to read your core memory registries, respect your design constraints, and hot-reload your active Dioxus UI workspace—all entirely in the background. ## 🎯 The Mission The MorBlogger UI is built to be a lightning-fast, offline-first desktop environment. Adding heavy HTTP servers or Python runtimes to the core app violates the suckless design philosophy. This MCP engine operates as an invisible, native Rust sub-process spawned by your AI client. It communicates with your AI via standard input/output (JSON-RPC) and reads the live state of your Dioxus UI via a hyper-efficient, zero-dependency Unix `/tmp` file drop. ## 🛠️ The Tools (Capabilities) When an AI connects to this engine, it gains access to the following tools: ### `read_theme_manifest` Queries the core compiler memory to return the exact CSS and JS dependencies for any template module (e.g., `mor` from the header registry). ### `get_module_constraints` Fetches the required structural rules, mandatory CSS classes (like `.panel-toggle`), and valid `{{VARIABLES}}` for a module so the AI never breaks your theme architecture. ### `write_module_xml` Safely overwrites the XML file for a specific workbench module. If the MorBlogger UI is running, this triggers an instant, mouse-free hot-reload of the preview canvas. ### `read_live_workspace` (Opt-in) Reads the live, unsaved state of the MorBlogger UI. If the user enables the AI Bridge, the MCP engine can read exactly which file is currently active in the UI and inspect unsaved keystrokes. ## 🚀 Installation & Wiring You do not run this binary manually. You compile it, and then hand the absolute path to your preferred AI client so it can spawn the server in the background. ### 1. Compile the Release Binary Clone the repository and build the optimized engine: ```bash cargo build --release ``` Your compiled binary will be located at: ```text target/release/mor-blogger-theme-editor-mcp ``` ### 2. Connect Your AI Client Provide the absolute path of your compiled binary to your AI environment. #### For Claude Desktop (Linux) Open or create the configuration file at `~/.config/Claude/claude_desktop_config.json` and add the engine to the `mcpServers` object: ```json { "mcpServers": { "mor_blogger_engine": { "command": "/absolute/path/to/your/target/release/mor-blogger-theme-editor-mcp", "args": [] } } } ``` **Note:** Restart Claude Desktop completely for the changes to take effect. #### For Cursor IDE 1. Open Cursor Settings. 2. Navigate to **Features > MCP**. 3. Click **+ Add New MCP Server**. 4. Name it `mor_blogger_engine`, set the type to `command`, and paste the absolute path to the binary. #### For Terminal CLI Agents (Grok / Antigravity) If you prefer a pure terminal workflow, attach the binary path directly via your agent's config file (usually located in `~/.config/`). **TOML Configuration Example:** ```toml [mcp.servers.mor_engine] command = "/absolute/path/to/your/target/release/mor-blogger-theme-editor-mcp" args = [] ``` **Command-Line Flag Example:** ```bash grok chat --mcp-server mor_engine="/absolute/path/to/your/target/release/mor-blogger-theme-editor-mcp" ``` ## Security Note The `read_live_workspace` tool requires the user to explicitly enable the **AI Bridge** toggle inside the MorBlogger app preferences. If this is toggled off, the engine remains securely blind to the live Dioxus environment.