Content
# prompt-toolkit-mcp



> Production-tested AI prompt templates for code review, debugging, testing, and architecture — available as an MCP server for Claude Code and other AI assistants.
Stop writing the same "act as a senior engineer and review my code" prompt from scratch every time. This MCP server gives your AI assistant 5 battle-tested prompt templates instantly.
---
## Available Prompts
| # | Name | Category | What it does |
|---|------|----------|--------------|
| 1 | Comprehensive Code Review | code-review | 6-dimension review: correctness, readability, error handling, security, performance, testability |
| 2 | Root Cause Analysis | debugging | 7-step diagnostic framework from stack traces to root cause + fix recommendation |
| 3 | Unit Test Generator | testing | Generates tests for happy paths, edge cases, error handling, security inputs, state transitions |
| 4 | System Design | architecture | Transforms requirements into a complete architecture doc: components, APIs, infra, trade-offs |
| 5 | README Generator | documentation | Creates professional READMEs that convert visitors into users in under 10 minutes |
---
## Quick Start
### Option 1: Claude Desktop / Claude Code
Add to your `claude_desktop_config.json` (macOS: `~/Library/Application Support/Claude/claude_desktop_config.json`):
```json
{
"mcpServers": {
"prompt-toolkit": {
"command": "uvx",
"args": ["prompt-toolkit-mcp"]
}
}
}
```
Restart Claude Desktop. You can now ask:
> "Use the code-review prompt to review this function: ..."
### Option 2: Run directly
```bash
uvx prompt-toolkit-mcp
```
### Option 3: Claude Code CLI
```bash
claude mcp add prompt-toolkit uvx prompt-toolkit-mcp
```
---
## MCP Tools
### `list_prompts()`
Returns a formatted table of all available prompts with categories, descriptions, and variable lists.
### `get_prompt(name)`
Returns the full prompt template with `{variable}` placeholders ready for you to fill in.
**Parameters:**
| Parameter | Type | Values |
|-----------|------|--------|
| `name` | string | `code-review`, `root-cause-analysis`, `unit-test-generator`, `system-design`, `readme-generator` |
**Example usage in Claude:**
> "Get the root-cause-analysis prompt"
### `analyze_code(code, review_type)`
Injects your code directly into the right prompt template and returns a ready-to-paste prompt.
**Parameters:**
| Parameter | Type | Default | Values |
|-----------|------|---------|--------|
| `code` | string | required | Your source code |
| `review_type` | string | `comprehensive` | `comprehensive`, `security`, `debugging`, `testing`, `architecture` |
**Example usage in Claude:**
> "Analyze this code for security issues: [paste code]"
---
## Usage Examples
### Code Review
In Claude, after installing the MCP server:
```
Get the code-review prompt, then fill in:
- {code}: [paste your code]
- {language}: Python 3.11 / FastAPI
- {component_description}: Payment webhook handler
- {constraints}: Must handle 500 RPS, 256MB Lambda
```
### Root Cause Analysis
```
Get the root-cause-analysis prompt, then fill in:
- {error_output}: [paste full stack trace]
- {application}: User auth service
- {recent_changes}: Deployed v2.3.1 yesterday
- {frequency}: ~200 errors/hour, started after deploy
```
### Quick Code Analysis
```
Use analyze_code with my function [paste code] — review_type: security
```
---
## Development Setup
```bash
git clone https://github.com/Money-Monkey-26/prompt-toolkit-mcp
cd prompt-toolkit-mcp
uv sync
uv run python server.py
```
Test the import:
```bash
uv run python -c "from server import mcp; print('OK')"
```
---
## Get the Full Toolkit
These 5 prompts are a sample from the **AI Developer Toolkit** — 32+ production-tested prompts covering every stage of the development lifecycle:
- Security audit, performance review, PR review, API design critique
- Incident post-mortem, log analysis, performance regression investigation
- Integration tests, E2E tests, property-based tests, test plans
- ADR (Architecture Decision Records), database schema review
- CONTRIBUTING guides, changelogs, technical blog posts
**Get the full AI Developer Toolkit ($39):** https://payhip.com/b/t172g
Landing page: https://money-monkey-26.github.io/ai-dev-toolkit/
---
## License
MIT