MCP Probe Kit

You can harness the MCP server to empower AI agents with full project context and delegated orchestration across the complete development lifecycle. This server provides structured, machine-friendly outputs and a suite of tools to analyze, design, implement, test, and release with AI-driven guidance.

How to use

To use this MCP server, connect it to an MCP client such as Cursor, Claude Desktop, Cline, Continue, or other MCP clients. You can run orchestration commands to automatically generate workflows, artifacts, and documentation. When you start a feature, bugfix, onboarding, UI work, product design, or Ralph loop, the server returns a structured execution plan and progressively creates artifacts like requirements, designs, and tests. Use the client to trigger tools step by step, persist generated files, and monitor progress through the provided progress notifications.

Key capabilities include structured output for easy parsing, real-time progress and cancellation support, graph context integration via GitNexus when available, and optional extensions for UI apps. If you need more context, you can request clarifications during requirements gathering, and then proceed with delegated orchestration to produce a complete set of artifacts and code.

How to install

Prerequisites: you need Node.js and npm installed on your system to run MCP server tooling. Ensure your environment can execute npm or npx commands from your shell.

Option A: Use directly with npx (recommended) for rapid access without installation.

Option B: Install globally for long-term use.

Additional setup notes

The server supports two local npx-based configurations shown in examples. One uses a basic npx invocation, and another uses an adjusted invocation that automatically accepts prompts. You can paste these configurations into your MCP client to enable the server.

You can also enable optional capabilities and UI app outputs via environment switches when running the server. For example, enabling extensions or UI data support can extend capabilities during tool execution.

Available tools

start_feature

Orchestrates end-to-end feature development from requirements to design and estimation

start_bugfix

Orchestrates bug analysis, fixing, and testing flows

start_onboard

Onboards a new project by generating context and setup docs

start_ui

Orchestrates UI design, component generation, and code output

start_product

Orchestrates product design from PRD to HTML prototype

start_ralph

Runs the Ralph Loop for iterative goal achievement

code_review

Analyzes code quality and suggests improvements

code_insight

Provides insights and context for code changes

fix_bug

Proposes and applies code fixes for defects

refactor

Suggests and applies refactoring to improve structure and maintainability

gencommit

Generates concise, professional commit messages

git_work_report

Generates work reports from Git history and diffs

gentest

Generates tests based on code and requirements

init_project

Initializes a new project with standard docs and structure

init_project_context

Generates project context documentation and conventions

add_feature

Adds a feature specification and related artifacts

estimate

Estimates effort and resources for a feature or task

interview

Conducts a structured requirements interview

ask_user

AI-driven proactive questioning to clarify requirements

ui_design_system

Generates a design system and UI tokens for consistency

ui_search

Searches UI/UX data using BM25 for rapid retrieval

sync_ui_data

Synchronizes latest UI/UX data for offline use