aicode-toolkit
Toolkit for Coding Agents to work reliably with repo of any size.
claude mcp add --transport stdio agiflow-aicode-toolkit npx -y @agiflowai/scaffold-mcp mcp-serve --admin-enable
How to use
AI Code Toolkit exposes a set of MCP servers that work together to provide scaffold generation, file-pattern guidance, and style/component discovery for code projects. The scaffold-mcp server handles boilerplate and feature scaffolding through commands like list-boilerplates and use-boilerplate, enabling rapid project setup. The architect-mcp server applies file-level design patterns and runs review checks against RULES.yaml to keep edits aligned with project conventions. The style-system server helps you discover themes, CSS classes, and components to ensure UI consistency across the codebase. To use them, configure the MCP with the provided commands, then interact with the agents (Claude Code, Cursor, Gemini CLI, etc.) to drive creation, editing, and review workflows using the built-in core and admin tools described in the repository documentation.
How to install
Prerequisites:
- Node.js ≥ 18 installed on your machine
- npm or yarn available
Install and configure the MCP servers:
-
Install the MCP provider suite (scaffold-mcp, architect-mcp, style-system) via npx as shown in the example configuration. These packages are published under @agiflowai scope.
-
Create an mcp configuration (see mcp_config example) in your project root, for example as .mcp.json or within your agent config, depending on your setup.
-
Start or verify the MCP servers using your agent or CLI workflow. For example, you can run the servers individually or in combination as part of your agent’s configuration.
-
Ensure your environment allows the MCP tooling to fetch templates and boilerplates from the npm registry (network access).
Additional notes
Tips and notes:
- You can enable admin tooling with --admin-enable to allow template authors and design-pattern/rule editors to manage content.
- When using design-pattern-tool, options like codex can filter and surface relevant design patterns for your codebase.
- The review-tool option lets you route review output through an LLM (e.g., Gemini CLI) to augment or automate feedback on edits.
- If you experience issues with boilerplate discovery, restart the agent or re-run the list-boilerplates and use-boilerplate sequences to refresh prompts.
- The repository exposes built-in templates for Next.js 15, TypeScript libraries, and MCP-packaged templates which may be referenced by the mcp-serve commands.
- Ensure you follow the Quick Start section of the README for precise usage patterns and examples specific to your agent and project type.
Related MCP Servers
obsidian -tools
Add Obsidian integrations like semantic search and custom Templater prompts to Claude or any MCP client.
interactive
Vibe coding should have human in the loop! interactive-mcp: Local, cross-platform MCP server for interact with your AI Agent
mcp
Octopus Deploy Official MCP Server
furi
CLI & API for MCP management
mcp -arangodb
This is a TypeScript-based MCP server that provides database interaction capabilities through ArangoDB. It implements core database operations and allows seamless integration with ArangoDB through MCP tools. You can use it wih Claude app and also extension for VSCode that works with mcp like Cline!
databutton
Databutton MCP Server