# Wassden Mcp

> **Note: Early development phase. stable production use is not yet possible.** `wassden` is an MCP-based toolkit that enforces spec-driven development for AI agents, inspired by kiro. Acting as a "warden," it guides agents to generate and follow structured specifications, serving as an assistant that ensures agents create

- **Type:** MCP server
- **Install:** `agentstack add mcp-tokusumi-wassden-mcp`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [tokusumi](https://agentstack.voostack.com/s/tokusumi)
- **Installs:** 0
- **Category:** [Integrations](https://agentstack.voostack.com/c/integrations)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [tokusumi](https://github.com/tokusumi)
- **Source:** https://github.com/tokusumi/wassden-mcp

## Install

```sh
agentstack add mcp-tokusumi-wassden-mcp
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

# wassden

[](https://github.com/tokusumi/wassden-mcp/actions/workflows/ci.yml)
[](https://www.python.org/)
[](https://opensource.org/licenses/MIT)
[](https://modelcontextprotocol.io/)

**Spec-Driven Development MCP server that transforms any LLM into a systematic development agent**

> 🚀 Generate requirements → design → tasks → code with 100% traceability and validation

## ⚡ Quick Start (MCP)

```bash
# Add to Claude Code
claude mcp add wassden -- uvx --from git+https://github.com/tokusumi/wassden-mcp wassden start-mcp-server --transport stdio

# Test it
uvx --from git+https://github.com/tokusumi/wassden-mcp wassden prompt-requirements --userInput "Create a TODO app"
```

## 🎯 What It Does

- ✅ **Requirements → Design → Tasks → Code** - Complete SDD workflow with structured prompts
- 🔒 **100% READ-ONLY** - No file modification risk, only analysis and prompt generation
- 🌍 **Auto Language Detection** - Seamless Japanese/English support
- ⚡ **Ultra-fast Performance** -  📚 For detailed setup and development installation, see [docs/development.md](docs/development.md)

## 🚀 Basic Usage

### 1. Analyze & Generate Requirements

```bash
# Analyze project description and generate requirements prompt
uvx --from git+https://github.com/tokusumi/wassden-mcp wassden prompt-requirements \
  --userInput "Create a task management API with authentication"
```

### 2. Validate Generated Specs

```bash
# Validate specifications with auto language detection
uvx --from git+https://github.com/tokusumi/wassden-mcp wassden validate-requirements specs/my-feature/requirements.md
uvx --from git+https://github.com/tokusumi/wassden-mcp wassden validate-design specs/my-feature/design.md  
uvx --from git+https://github.com/tokusumi/wassden-mcp wassden validate-tasks specs/my-feature/tasks.md
```

### 3. Get Traceability Matrix

```bash
# Generate complete dependency mapping
uvx --from git+https://github.com/tokusumi/wassden-mcp wassden get-traceability
```

## 🛠️ Core Tools

### Requirements Generation
- **`prompt-requirements`** - Analyzes input for completeness and generates EARS requirements prompt
  - Default: Checks completeness, asks questions if incomplete, generates prompt if complete
  - `--force`: Skip completeness verification and generate requirements prompt

### Design & Planning  
- **`prompt-design`** - Generates architectural design prompt from requirements
- **`prompt-tasks`** - Creates WBS task breakdown prompt from design

### Validation Suite
- **`validate-requirements`** - Ensures EARS format and completeness
- **`validate-design`** - Checks architectural consistency
- **`validate-tasks`** - Verifies task dependencies and coverage

### Traceability & Analysis
- **`get-traceability`** - Complete REQ↔DESIGN↔TASK dependency mapping
- **`analyze-changes`** - Impact analysis for specification changes
- **`generate-review-prompt`** - Task-specific implementation review

> 📚 See [full tool documentation](docs/tools.md) for detailed usage and examples

## 📊 Performance

- **Response Time**:  🧪 Run benchmarks: `python benchmarks/run_all.py`

## 🧪 Development Mode Features (Experimental)

**⚠️ NOTE: Experimental features for internal validation and benchmarking. Requires dev installation.**

The experiment framework provides validation and benchmarking capabilities for the wassden toolkit itself. These features are only available in development mode:

```bash
# Install with dev dependencies
git clone https://github.com/tokusumi/wassden-mcp
cd wassden-mcp
uv sync  # Installs all dependencies including dev extras

# Run experiments (dev mode only)
uv run wassden experiment run ears        # EARS pattern validation
uv run wassden experiment run performance # Performance benchmarking  
uv run wassden experiment run language    # Language detection testing
```

### Key Capabilities

- **Statistical Analysis**: Mean, variance, confidence intervals (REQ-04)
- **Experiment Management**: Save/load/compare configurations (REQ-05, REQ-06)
- **Comparative Analysis**: Statistical significance testing (REQ-07)
- **Resource Constraints**: 10-minute timeout, 100MB memory limit (NFR-01, NFR-02)

> 📚 See [Experiment Framework Documentation](docs/experiment-framework.md) for detailed usage

## 🎯 Use Cases

- **Development Teams**: Systematic requirements gathering and project planning
- **AI Agents**: Structured prompts for complex development workflows
- **Technical Writers**: Automated documentation validation and traceability
- **Quality Assurance**: Built-in validation with actionable feedback

## 📚 Documentation

- **[Getting Started](docs/quickstart.md)** - Installation and first steps
- **[Tool Reference](docs/tools.md)** - Complete tool documentation
- **[Spec Format Guide](docs/spec-format.md)** - Requirements, design, and tasks format
- **[Validation Rules](docs/validation/)** - EARS format and traceability requirements
- **[Validation System](docs/validation.md)** - Comprehensive validation with AST-based analysis
- **[Development](docs/development.md)** - Development setup and contributing
- **[Experiment Framework](docs/experiment-framework.md)** - Experimental validation features (dev only)
- **[Examples](docs/ja/spec-example/)** - Sample specifications (Japanese/English)

## 🤝 Contributing

Contributions are welcome! See [docs/development.md](docs/development.md) for development setup.

```bash
make check  # Run all quality checks before committing
```

## 📄 License

MIT License - see [LICENSE](LICENSE) file for details.

---

**Built with ❤️ for the AI-driven development community**

## Source & license

This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [tokusumi](https://github.com/tokusumi)
- **Source:** [tokusumi/wassden-mcp](https://github.com/tokusumi/wassden-mcp)
- **License:** MIT

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/mcp-tokusumi-wassden-mcp
- Seller: https://agentstack.voostack.com/s/tokusumi
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.