Probe Kit MCP Server

mcp-probe-kit — Know the Context, Feed the Moment

---

Talk is cheap, show me the Context.

mcp-probe-kit is a protocol-level toolkit designed for developers who want AI to truly understand their project's intent. It's not just a collection of 29 tools—it's a context-aware system that helps AI agents grasp what you're building.

Languages: English | 简体中文 | 日本語 | 한국어 | Español | Français | Deutsch | Português (BR)

🚀 AI-Powered Complete Development Toolkit - Covering the Entire Development Lifecycle

A powerful MCP (Model Context Protocol) server providing 29 tools covering the complete workflow from product analysis to final release (Requirements → Design → Development → Quality → Release), all tools support structured output.

🎉 v3.0 Major Update: Streamlined tool count, focus on core competencies, eliminate choice paralysis, let AI do more native work

Supports All MCP Clients: Cursor, Claude Desktop, Cline, Continue, and more

Protocol Version: MCP 2025-11-25 · SDK: @modelcontextprotocol/sdk 1.27.1

---

📚 Complete Documentation

👉 https://mcp-probe-kit.bytezonex.com

• Quick Start - Setup in 5 minutes
• Local Memory Stack (Qdrant + Nomic Embed) - Docker Compose, ports 50008 / 50012, MCP env
• All Tools - Complete list of 29 tools
• Best Practices - Full development workflow guide
• v3.0 Migration Guide - Upgrade from v2.x to v3.0

---

✨ Core Features

📦 28 Tools

• 🔄 Workflow Orchestration (6 tools) - One-click complex development workflows
  • start_feature, start_bugfix, start_onboard, start_ui, start_product, start_ralph
• 🔍 Code Analysis (4 tools) - Code quality, refactoring, and graph insight
  • code_review, code_insight, fix_bug, refactor
• 📝 Git Tools (2 tools) - Git commits and work reports
  • gencommit, git_work_report
• ⚡ Code Generation (1 tool) - Test generation
  • gentest
• 📦 Project Management (6 tools) - Project initialization and requirements management
  • init_project, init_project_context, add_feature, estimate, interview, ask_user
• 🎨 UI/UX Utilities (3 tools) - Design systems and UI data synchronization
  • ui_design_system, ui_search, sync_ui_data
• 🧠 Memory & Cursor History (7 tools) - Reusable asset memory and local Cursor conversation retrieval
  • search_memory, read_memory_asset, memorize_asset, scan_and_extract_patterns, cursor_list_conversations, cursor_search_conversations, cursor_read_conversation

🧠 Code Graph Bridge (GitNexus)

• code_insight bridges GitNexus by default for query/context/impact analysis
• The bridge launches npx -y gitnexus@latest mcp by default to reduce stale package risk
• init_project_context bootstraps baseline graph docs under docs/graph-insights/; if docs/project-context.md already exists, it preserves the old context docs and only backfills graph docs plus the index entry
• start_feature refreshes the GitNexus index and runs task-level query/context/impact narrowing before spec generation to reduce over-scoping
• start_bugfix refreshes the GitNexus index and runs task-level graph analysis before TBP RCA to constrain failure boundary and blast radius
• Older projects that already have project-context.md but no graph docs are bootstrapped automatically through the init_project_context step
• If GitNexus is unavailable, the server falls back automatically without breaking orchestration
• Real graph queries read the .gitnexus index; docs/graph-insights/latest.md|json are readable snapshots for humans and AI agents
• Graph snapshots are exposed as resources (probe://graph/latest, probe://graph/history, probe://graph/latest.md)
• Graph snapshots are also persisted to readable files in .mcp-probe-kit/graph-snapshots (customizable via MCP_GRAPH_SNAPSHOT_DIR)
• Tool responses include _meta.graph with snapshot URI and local JSON/Markdown file paths

🐛 TBP 8-Step RCA for Bug Workflows

• start_bugfix defaults to Toyota-style TBP 8-step root-cause analysis before repair
• fix_bug returns a structured TBP skeleton covering phenomenon, timeline, ruled-out paths, boundary, root cause, evidence, and repair plan
• This makes bug, regression, anomaly, and "why didn't it work" investigations follow analyze-first discipline instead of patching symptoms

🧠 Memory Retrieval and Cursor History

• Memory tools use Qdrant as the vector database backend
• Embedding service supports two modes:
  • ollama
  • openai-compatible
• Cursor history tools read the local Cursor database directly through Node.js, without Python bridge
• Cursor history currently supports:
  • Windows: %APPDATA%\\Cursor\\User\\globalStorage\\state.vscdb
  • macOS: ~/Library/Application Support/Cursor/User/globalStorage/state.vscdb
  • Linux: ~/.config/Cursor/User/globalStorage/state.vscdb

Memory tools:

• search_memory - Semantic search across the shared memory pool (optionally prefer type / tags)
• memorize_asset - Persist reusable code/spec/pattern assets into vector memory
• read_memory_asset - Read full asset content by asset_id
• scan_and_extract_patterns - Extract reusable patterns from code/file/directory before deciding whether to persist

Cross-repo memory pools: do not rely on source_project / source_path for shared retrieval; put file paths in content instead. Search injection hides foreign sourcePath unless MEMORY_REPO_ID matches or MEMORY_SEARCH_SHOW_SOURCE=true.

Memory backend and embedding configuration:

• Vector database: Qdrant
• Recommended local setup: Qdrant (port 50008) + Infinity / nomic-embed (port 50012) — lighter than Ollama; see Local Memory Stack guide (中文: memory-local-setup.zh-CN.md)
• Supported embedding providers:
  • ollama
  • openai-compatible (Infinity, OpenAI, etc.)
• Required environment variables for memory write/search:
  • MEMORY_QDRANT_URL
  • MEMORY_EMBEDDING_URL
  • MEMORY_EMBEDDING_MODEL
• Optional environment variables:
  • MEMORY_QDRANT_API_KEY
  • MEMORY_QDRANT_COLLECTION (default: mcp_probe_memory)
  • MEMORY_EMBEDDING_API_KEY
  • MEMORY_EMBEDDING_PROVIDER (ollama by default)
  • MEMORY_SEARCH_LIMIT (default: 3)
  • MEMORY_SUMMARY_MAX_CHARS (default: 280)
  • MEMORY_SEARCH_MIN_SCORE (default: 0 = disabled; try 0.72 for noisy pools)
  • MEMORY_SEARCH_SHOW_SOURCE (default: false)
  • MEMORY_REPO_ID (optional; show sourcePath only when sourceProject matches)
  • MEMORY_INJECTION_CONTENT_MAX_CHARS (default: 1500; max content per hit injected into start_* guides)
• Behavior notes:
  • Read-only memory access only requires MEMORY_QDRANT_URL
  • Memory write is enabled only when MEMORY_QDRANT_URL, MEMORY_EMBEDDING_URL, and MEMORY_EMBEDDING_MODEL are all configured
  • The Qdrant collection is auto-created on first write, and vector dimension is inferred from the first embedding response

Recommended local memory setup (Qdrant + Nomic Embed / Infinity):

Full Docker Compose, ports, and troubleshooting: docs/memory-local-setup.md

{
  "mcpServers": {
    "mcp-probe-kit": {
      "command": "npx",
      "args": ["-y", "mcp-probe-kit@latest"],
      "env": {
        "MEMORY_QDRANT_URL": "http://127.0.0.1:50008",
        "MEMORY_QDRANT_API_KEY": "your-qdrant-api-key",
        "MEMORY_QDRANT_COLLECTION": "mcp_probe_memory",
        "MEMORY_EMBEDDING_PROVIDER": "openai-compatible",
        "MEMORY_EMBEDDING_URL": "http://127.0.0.1:50012/embeddings",
        "MEMORY_EMBEDDING_MODEL": "nomic-ai/nomic-embed-text-v1.5",
        "MEMORY_EMBEDDING_API_KEY": "your-infinity-api-key",
        "MEMORY_SEARCH_LIMIT": "3",
        "MEMORY_SUMMARY_MAX_CHARS": "280"
      }
    }
  }
}

Alternative: Qdrant + Ollama (if you already run Ollama):

docker run -d --name mcp-qdrant -p 6333:6333 qdrant/qdrant
ollama pull nomic-embed-text

"MEMORY_QDRANT_URL": "http://127.0.0.1:6333",
"MEMORY_EMBEDDING_PROVIDER": "ollama",
"MEMORY_EMBEDDING_URL": "http://127.0.0.1:11434/api/embeddings",
"MEMORY_EMBEDDING_MODEL": "nomic-embed-text"

OpenAI-compatible embedding (hosted API):

{
  "mcpServers": {
    "mcp-probe-kit": {
      "command": "npx",
      "args": ["-y", "mcp-probe-kit@latest"],
      "env": {
        "MEMORY_QDRANT_URL": "http://127.0.0.1:6333",
        "MEMORY_QDRANT_COLLECTION": "mcp_probe_memory",
        "MEMORY_EMBEDDING_PROVIDER": "openai-compatible",
        "MEMORY_EMBEDDING_URL": "https://your-embedding-endpoint/v1/embeddings",
        "MEMORY_EMBEDDING_API_KEY": "your-api-key",
        "MEMORY_EMBEDDING_MODEL": "text-embedding-3-small"
      }
    }
  }
}

Cursor history tools:

• cursor_list_conversations - List recent local Cursor conversations by title/workspace
• cursor_search_conversations - Search local Cursor history by keyword or request id
• cursor_read_conversation - Read a single local Cursor conversation timeline by composer_id

🎯 Structured Output

Core and orchestration tools support structured output, returning machine-readable JSON data, improving AI parsing accuracy, supporting tool chaining and state tracking.

⏱️ Native Tasks, Progress, and Cancellation

• Built on MCP SDK native task support (taskStore + taskMessageQueue)
• Supports task lifecycle endpoints: tasks/get, tasks/result, tasks/list, tasks/cancel
• Advertises capabilities.tasks.requests.tools.call so clients can create tasks for tools/call
• Emits notifications/progress when client provides _meta.progressToken
• Handles request cancellation via AbortSignal and returns a clear cancellation error
• Long-running orchestration tools (start_*) and sync_ui_data support cooperative cancellation/progress callbacks

🔌 Extensions & UI Apps (Optional)

• Trace metadata passthrough: request _meta.trace is preserved in tool responses (_meta.trace)
• Optional extensions capability switch: enable with MCP_ENABLE_EXTENSIONS_CAPABILITY=1
• Optional MCP Apps resource output for UI tools: enable with MCP_ENABLE_UI_APPS=1
• UI tools can expose preview resources via ui://... and response _meta.ui.resourceUri

🧭 Delegated Orchestration Protocol

All start_* orchestration tools return an execution plan in structuredContent.metadata.plan.
AI needs to call tools step by step and persist files, rather than the tool executing internally.

Plan Schema (Core Fields):

{
  "mode": "delegated",
  "steps": [
    {
      "id": "spec",
      "tool": "add_feature",
      "args": { "feature_name": "user-auth", "description": "User authentication feature" },
      "outputs": ["docs/specs/user-auth/requirements.md"]
    }
  ]
}

Field Description:

• mode: Fixed as delegated
• steps: Array of execution steps
• tool: Tool name (e.g. add_feature)
• action: Manual action description when no tool (e.g. update_project_context)
• args: Tool parameters
• outputs: Expected artifacts
• when/dependsOn/note: Optional conditions and notes

🧩 Structured Output Field Specification (Key Fields)

Both orchestration and atomic tools return structuredContent, common fields:

• summary: One-line summary
• status: Status (pending/success/failed/partial)
• steps: Execution steps (orchestration tools)
• artifacts: Artifact list (path + purpose)
• metadata.plan: Delegated execution plan (only start_*)
• specArtifacts: Specification artifacts (start_feature)
• estimate: Estimation results (start_feature / estimate)

🧠 Requirements Clarification Mode (Requirements Loop)

When requirements are unclear, use requirements_mode=loop in start_feature / start_bugfix / start_ui.
This mode performs 1-2 rounds of structured clarification before entering spec/fix/UI execution.

Example:

{
  "feature_name": "user-auth",
  "description": "User authentication feature",
  "requirements_mode": "loop",
  "loop_max_rounds": 2,
  "loop_question_budget": 5
}

🧩 Template System (Regular Model Friendly)

add_feature supports template profiles, default auto auto-selects: prefers guided when requirements are incomplete (includes detailed filling rules and checklists), selects strict when requirements are complete (more compact structure, suitable for high-capability models or archival scenarios).

Example:

{
  "description": "Add user authentication feature",
  "template_profile": "auto"
}

Applicable Tools:

• start_feature passes template_profile to add_feature
• start_bugfix / start_ui also support template_profile for controlling guidance strength (auto/guided/strict)

Template Profile Strategy:

• guided: Less/incomplete requirements info, regular model priority
• strict: Requirements structured, prefer more compact guidance
• auto: Default recommendation, auto-selects guided/strict

🔄 Workflow Orchestration

6 intelligent orchestration tools that automatically combine multiple basic tools for one-click complex development workflows:

• start_feature - New feature development (Requirements → Design → Estimation)
• start_bugfix - Bug fixing (TBP 8-step RCA → Fix → Testing)
• start_onboard - Project onboarding (Generate project context docs)
• start_ui - UI development (Design system → Components → Code)
• start_product - Product design (PRD → Prototype → Design system → HTML)
• start_ralph - Ralph Loop (Iterative development until goal completion)

🚀 Product Design Workflow

start_product is a complete product design orchestration tool, from requirements to interactive prototype:

Workflow:

1. Requirements Analysis - Generate standard PRD (product overview, feature requirements, page list)
2. Prototype Design - Generate detailed prototype docs for each page
3. Design System - Generate design specifications based on product type
4. HTML Prototype - Generate interactive prototype viewable in browser
5. Project Context - Auto-update project documentation

Structured Output Additions:

• start_product.structuredContent.artifacts: Artifact list (PRD, prototypes, design system, etc.)
• interview.structuredContent.mode: usage / questions / record

🎨 UI/UX Pro Max

4 UI/UX tools with start_ui as the unified entry point:

• start_ui - One-click UI development (supports intelligent mode) (orchestration tool)
• ui_design_system - Intelligent design system generation
• ui_search - UI/UX data search (BM25 algorithm)
• sync_ui_data - Sync latest UI/UX data locally

Note: start_ui automatically calls ui_design_system and ui_search, you don't need to call them separately.

Inspiration:

• ui-ux-pro-max-skill - UI/UX design system philosophy
• json-render - JSON template rendering engine

Skill Bridge for UI/PRD workflows:

• start_ui and start_product now include a Skill Bridge section in guidance and structuredContent.metadata.skills.
• Recommended skill call order: ui-ux-pro-max → interaction-design → frontend-design.
• If some skills are missing, workflow continues with MCP main plan and marks unavailable skills in metadata.

Why use sync_ui_data?

Our start_ui tool relies on a rich UI/UX database (colors, icons, charts, components, design patterns, etc.) to generate high-quality design systems and code. This data comes from npm package uipro-cli, including:

• 🎨 Color schemes (mainstream brand colors, color palettes)
• 🔣 Icon libraries (React Icons, Heroicons, etc.)
• 📊 Chart components (Recharts, Chart.js, etc.)
• 🎯 Landing page templates (SaaS, e-commerce, government, etc.)
• 📐 Design specifications (spacing, fonts, shadows, etc.)

Data Sync Strategy:

1. Embedded Data: Synced at build time, works offline
2. Background Auto Sync: Downloads latest data to ~/.mcp-probe-kit/ui-ux-data/ without changing current session output
3. Next-Start Activation: Newly downloaded data is applied on next process start (keeps current session deterministic)
4. Manual Sync: Use sync_ui_data to force refresh cache immediately (still applies next start by default)

This ensures start_ui can generate professional-grade UI code even offline.

🎤 Requirements Interview

2 interview tools to clarify requirements before development:

• interview - Structured requirements interview
• ask_user - AI proactive questioning

---

🧭 Tool Selection Guide

When to use orchestration tools vs individual tools?

Use orchestration tools (start_*) when:

• ✅ Need complete workflow (multiple steps)
• ✅ Want to automate multiple tasks
• ✅ Need to generate multiple artifacts (docs, code, tests, etc.)

Use individual tools when:

• ✅ Only need specific functionality
• ✅ Already have project context docs
• ✅ Need more fine-grained control

Common Scenario Selection

---

🚀 Quick Start

Method 1: Use directly with npx (Recommended)

No installation needed, use the latest version directly.

Cursor / Cline Configuration

Config file location:

• Windows: %APPDATA%\Cursor\User\globalStorage\saoudrizwan.claude-dev\settings\cline_mcp_settings.json
• macOS: ~/Library/Application Support/Cursor/User/globalStorage/saoudrizwan.claude-dev/settings/cline_mcp_settings.json
• Linux: ~/.config/Cursor/User/globalStorage/saoudrizwan.claude-dev/settings/cline_mcp_settings.json

Config content:

{
  "mcpServers": {
    "mcp-probe-kit": {
      "command": "npx",
      "args": ["mcp-probe-kit@latest"]
    }
  }
}

Claude Desktop Configuration

Config file location:

• Windows: %APPDATA%\Claude\claude_desktop_config.json
• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Linux: ~/.config/Claude/claude_desktop_config.json

Config content:

{
  "mcpServers": {
    "mcp-probe-kit": {
      "command": "npx",
      "args": ["-y", "mcp-probe-kit@latest"]
    }
  }
}

OpenCode Configuration

Config file location:

• Project-level: opencode.json (in project root)
• Global: ~/.config/opencode/opencode.json

Config content:

{
  "mcp": {
    "mcp-probe-kit": {
      "type": "local",
      "command": ["npx", "-y", "mcp-probe-kit@latest"],
      "enabled": true
    }
  }
}

Note: OpenCode uses opencode.json with a different schema from Cursor/Claude Desktop. The key mcp replaces mcpServers, command is an array, type: "local" is required, and environment variables use environment instead of env. See OpenCode MCP docs for details.

Method 2: Global Installation

npm install -g mcp-probe-kit

Use in config file:

{
  "mcpServers": {
    "mcp-probe-kit": {
      "command": "mcp-probe-kit"
    }
  }
}

Optional Memory System Setup

If you want to use memorize_asset, read_memory_asset, and scan_and_extract_patterns, you need both:

1. A Qdrant vector database
2. An embedding service in either ollama or openai-compatible mode

Full guide (Docker Compose for Qdrant + Infinity, ports 50008 / 50012, MCP env, smoke tests):

• English: docs/memory-local-setup.md
• 中文: docs/memory-local-setup.zh-CN.md

Option A: Qdrant + Nomic Embed / Infinity (recommended)

Lightweight local stack; no Ollama. Deploy Qdrant and nomic-embed via Docker Compose (see guide), then:

{
  "mcpServers": {
    "mcp-probe-kit": {
      "command": "npx",
      "args": ["-y", "mcp-probe-kit@latest"],
      "env": {
        "MEMORY_QDRANT_URL": "http://127.0.0.1:50008",
        "MEMORY_QDRANT_API_KEY": "your-qdrant-api-key",
        "MEMORY_QDRANT_COLLECTION": "mcp_probe_memory",
        "MEMORY_EMBEDDING_PROVIDER": "openai-compatible",
        "MEMORY_EMBEDDING_URL": "http://127.0.0.1:50012/embeddings",
        "MEMORY_EMBEDDING_MODEL": "nomic-ai/nomic-embed-text-v1.5",
        "MEMORY_EMBEDDING_API_KEY": "your-infinity-api-key",
        "MEMORY_SEARCH_LIMIT": "3",
        "MEMORY_SUMMARY_MAX_CHARS": "280"
      }
    }
  }
}

Embedding URL must be /embeddings (not /v1/embeddings). Qdrant requires api-key when QDRANT__SERVICE__API_KEY is set.

Option B: Qdrant + Ollama

docker run -d --name mcp-qdrant -p 6333:6333 qdrant/qdrant
ollama pull nomic-embed-text

"MEMORY_QDRANT_URL": "http://127.0.0.1:6333",
"MEMORY_EMBEDDING_PROVIDER": "ollama",
"MEMORY_EMBEDDING_URL": "http://127.0.0.1:11434/api/embeddings",
"MEMORY_EMBEDDING_MODEL": "nomic-embed-text"

Option C: Qdrant + hosted OpenAI-compatible API

"MEMORY_QDRANT_URL": "http://127.0.0.1:50008",
"MEMORY_EMBEDDING_PROVIDER": "openai-compatible",
"MEMORY_EMBEDDING_URL": "https://your-embedding-endpoint/v1/embeddings",
"MEMORY_EMBEDDING_API_KEY": "your-api-key",
"MEMORY_EMBEDDING_MODEL": "text-embedding-3-small"

Memory Environment Variables

• MEMORY_QDRANT_URL: Qdrant base URL, required for all memory features
• MEMORY_QDRANT_API_KEY: Optional Qdrant API key
• MEMORY_QDRANT_COLLECTION: Collection name, default mcp_probe_memory
• MEMORY_EMBEDDING_PROVIDER: ollama or openai-compatible
• MEMORY_EMBEDDING_URL: Embedding endpoint URL
• MEMORY_EMBEDDING_API_KEY: Optional for Ollama, usually required for hosted OpenAI-compatible providers
• MEMORY_EMBEDDING_MODEL: Default is nomic-embed-text
• MEMORY_SEARCH_LIMIT: Default search result count is 3
• MEMORY_SUMMARY_MAX_CHARS: Default summary truncation length is 280

Notes

• Memory write capability is enabled only when MEMORY_QDRANT_URL, MEMORY_EMBEDDING_URL, and MEMORY_EMBEDDING_MODEL are configured
• Memory read capability only requires MEMORY_QDRANT_URL
• Qdrant collections are auto-created on first write with Cosine distance
• Vector size is inferred from the first embedding response

Cursor History Support

Cursor local history tools do not require Qdrant or embedding configuration.

Supported platforms:

• Windows
• macOS
• Linux

Requirements:

• Cursor must be installed locally
• Cursor local database must exist under User/globalStorage/state.vscdb
• Compatibility depends on Cursor's current local database schema

Windows Notes for Graph Tools

Applies to code_insight, start_feature, start_bugfix, and init_project_context.

• The GitNexus bridge uses npx -y gitnexus@latest mcp by default.
• On Windows, the first cold start can take 20+ seconds because npx may check/download packages.
• Some GitNexus dependencies use tree-sitter-* native modules. If your machine lacks Visual Studio Build Tools, the first install may fail with errors like gyp ERR! find VS could not find a version of Visual Studio 2017 or newer to use.

Recommended on Windows:

1. Install Visual Studio Build Tools with the C++ workload if you use graph-aware tools regularly.
2. Prefer stable local/global CLI usage for GitNexus when your MCP client supports env.
3. Increase GitNexus connect/call timeouts on slower or first-run environments.

Quick install command (Windows):

winget install Microsoft.VisualStudio.2022.BuildTools

Example config using a preinstalled gitnexus CLI:

{
  "mcpServers": {
    "mcp-probe-kit": {
      "command": "mcp-probe-kit",
      "env": {
        "MCP_GITNEXUS_COMMAND": "gitnexus",
        "MCP_GITNEXUS_ARGS": "mcp",
        "MCP_GITNEXUS_CONNECT_TIMEOUT_MS": "30000",
        "MCP_GITNEXUS_TIMEOUT_MS": "45000"
      }
    }
  }
}

Restart Client

After configuration, completely quit and reopen your MCP client.

👉 Detailed Installation Guide

---

💡 Usage Examples

Daily Development

code_review @feature.ts    # Code review
gentest @feature.ts         # Generate tests
gencommit                   # Generate commit message

New Feature Development

start_feature user-auth "User authentication feature"
# Auto-complete: Requirements analysis → Design → Effort estimation

Bug Fixing

start_bugfix
# Then paste error message
# Auto-complete: Problem location → Fix solution → Test code

Product Design

start_product "Online Education Platform" --product_type=SaaS
# Auto-complete: PRD → Prototype → Design system → HTML prototype

UI Development

start_ui "Login Page" --mode=auto
# Auto-complete: Design system → Component generation → Code output

Project Context Documentation

# Single file mode (default) - Generate a complete project-context.md
init_project_context

# Modular mode - Generate 6 category docs (suitable for large projects)
init_project_context --mode=modular
# Generates: project-context.md (index) + 5 category docs

Git Work Report

# Generate daily report
git_work_report --date 2026-02-03

# Generate weekly report
git_work_report --start_date 2026-02-01 --end_date 2026-02-07

# Save to file
git_work_report --date 2026-02-03 --output_file daily-report.md
# Auto-analyze Git diff, generate concise professional report
# If direct command fails, auto-provides temp script solution (auto-deletes after execution)

👉 More Usage Examples

---

❓ FAQ

Q1: Tool not working or errors?

Check detailed logs:

Windows (PowerShell):

npx -y mcp-probe-kit@latest 2>&1 | Tee-Object -FilePath .\mcp-probe-kit.log

macOS/Linux:

npx -y mcp-probe-kit@latest 2>&1 | tee ./mcp-probe-kit.log

Q2: Client not recognizing tools after configuration?

1. Restart client (completely quit then reopen)
2. Check config file path is correct
3. Confirm JSON format is correct, no syntax errors
4. Check client developer tools or logs for error messages

Q3: How to update to latest version?

npx method (Recommended): Use @latest tag in config, automatically uses latest version.

Global installation method:

npm update -g mcp-probe-kit

Q4: Why are graph-aware tools slow or timing out on Windows the first time?

This usually affects code_insight, start_feature, start_bugfix, and init_project_context.

Common causes:

1. npx -y gitnexus@latest mcp performs a cold start and may spend 20+ seconds checking/downloading packages.
2. GitNexus may need native tree-sitter-* modules, which can require Visual Studio Build Tools on Windows.

If you see logs like:

gyp ERR! find VS could not find a version of Visual Studio 2017 or newer to use
gyp ERR! find VS - missing any VC++ toolset

Try this:

1. Install Visual Studio Build Tools with the C++ workload.
2. Retry once after dependencies finish installing.
3. If your client supports env, switch the bridge to a preinstalled gitnexus CLI and raise: MCP_GITNEXUS_CONNECT_TIMEOUT_MS MCP_GITNEXUS_TIMEOUT_MS

👉 More FAQ

---

🤝 Contributing

Issues and Pull Requests welcome!

Improvement suggestions:

• Add useful tools
• Optimize existing tool prompts
• Improve documentation and examples
• Fix bugs

---

📄 License

MIT License

---

🔗 Related Links

• Author: Kyle (小墨)
• GitHub: mcp-probe-kit
• npm: mcp-probe-kit
• Documentation: https://mcp-probe-kit.bytezonex.com

Related Projects:

• Model Context Protocol (MCP) - Official MCP protocol docs
• GitHub Spec-Kit - GitHub spec-driven development toolkit
• ui-ux-pro-max-skill - UI/UX design system philosophy source
• json-render - JSON template rendering engine inspiration
• uipro-cli - UI/UX data source

---

Made with ❤️ for AI-Powered Development