# Tool Reference
Complete documentation for all wassden MCP tools.
## Overview
wassden provides a comprehensive set of tools for Spec-Driven Development (SDD) workflows. All tools support automatic language detection (Japanese/English) and can be used via CLI or MCP integration.
## Prompt Generation Tools
These tools provide structured prompts for AI agents to create high-quality specifications.
### prompt-requirements
Analyzes user input for completeness and generates requirements prompts.
**Usage:**
```bash
# Default behavior - checks completeness first
uvx --from git+https://github.com/tokusumi/wassden-mcp wassden prompt-requirements \
--userInput "Create a task management API with authentication"
# Skip completeness verification
uvx --from git+https://github.com/tokusumi/wassden-mcp wassden prompt-requirements \
--userInput "Create a task management API with authentication" \
--force
```
**Parameters:**
- `userInput` (required): Project description or requirements
- `--force` (optional): Skip completeness verification before generating prompt
- `--language` (optional): Force language (ja/en), auto-detects by default
**Output:**
- Default: If incomplete, shows clarifying questions; if complete, shows structured prompt
- With --force: Skips completeness verification and generates structured prompt for EARS-format requirements
### prompt-design
Generates architectural design prompt from requirements.
**Usage:**
```bash
uvx --from git+https://github.com/tokusumi/wassden-mcp wassden prompt-design \
--requirementsPath specs/requirements.md
```
**Parameters:**
- `requirementsPath`: Path to requirements document
- Auto-detects language from document content
### prompt-tasks
Generates WBS task breakdown prompt from design.
**Usage:**
```bash
uvx --from git+https://github.com/tokusumi/wassden-mcp wassden prompt-tasks \
--designPath specs/design.md
```
**Parameters:**
- `designPath`: Path to design document
- Auto-detects language from document content
### prompt-code
Generates implementation prompt from all specifications.
**Usage:**
```bash
uvx --from git+https://github.com/tokusumi/wassden-mcp wassden prompt-code \
--requirementsPath specs/requirements.md \
--designPath specs/design.md \
--tasksPath specs/tasks.md
```
**Parameters:**
- `requirementsPath`: Path to requirements
- `designPath`: Path to design
- `tasksPath`: Path to tasks
### generate-review-prompt
Generates task-specific implementation review prompt.
**Usage:**
```bash
uvx --from git+https://github.com/tokusumi/wassden-mcp wassden generate-review-prompt TASK-01-01
```
**Parameters:**
- `taskId`: Specific task ID to review
- `requirementsPath` (optional): Custom requirements path
- `designPath` (optional): Custom design path
- `tasksPath` (optional): Custom tasks path
**Features:**
- Quality guardrails preventing test tampering
- Project-specific quality checks
- Requirements traceability validation
- Pass/fail criteria with actionable feedback
## Validation Tools
These tools validate agent-generated documents for quality and consistency.
### validate-requirements
Validates requirements document format and completeness.
**Usage:**
```bash
uvx --from git+https://github.com/tokusumi/wassden-mcp wassden validate-requirements \
specs/requirements.md
```
**Validation Checks:**
- EARS format compliance
- REQ-ID sequential numbering
- TR-ID test requirement numbering
- Section structure completeness
- Language consistency
### validate-design
Validates design document structure and references.
**Usage:**
```bash
uvx --from git+https://github.com/tokusumi/wassden-mcp wassden validate-design \
specs/design.md
```
**Validation Checks:**
- Component structure
- Interface definitions
- Requirements coverage (all REQs referenced)
- Design pattern consistency
### validate-tasks
Validates task dependencies and coverage.
**Usage:**
```bash
uvx --from git+https://github.com/tokusumi/wassden-mcp wassden validate-tasks \
specs/tasks.md
```
**Validation Checks:**
- Task ID format
- Dependency DAG (no circular dependencies)
- Requirements coverage (100% REQ + TR)
- Design component coverage
- Test scenario references (DC field)
## Traceability & Analysis Tools
These tools provide project insights and dependency tracking.
### get-traceability
Generates complete traceability matrix.
**Usage:**
```bash
uvx --from git+https://github.com/tokusumi/wassden-mcp wassden get-traceability
```
**Parameters:**
- `requirementsPath` (optional): Custom requirements path
- `designPath` (optional): Custom design path
- `tasksPath` (optional): Custom tasks path
**Output:**
```json
{
"requirements_to_design": {
"REQ-01": ["COMP-01", "COMP-02"],
"REQ-02": ["COMP-03"]
},
"requirements_to_tasks": {
"REQ-01": ["TASK-01-01", "TASK-01-02"],
"REQ-02": ["TASK-02-01"]
},
"design_to_tasks": {
"COMP-01": ["TASK-01-01"],
"COMP-02": ["TASK-01-02"]
},
"test_requirements_to_tasks": {
"TR-01": ["TASK-03-01"],
"TR-02": ["TASK-03-02"]
},
"coverage": {
"requirements": 100,
"design": 100,
"test_requirements": 100
}
}
```
### analyze-changes
Analyzes impact of specification changes.
**Usage:**
```bash
uvx --from git+https://github.com/tokusumi/wassden-mcp wassden analyze-changes \
--changedFile specs/requirements.md \
--changeDescription "Added authentication requirement REQ-10"
```
**Parameters:**
- `changedFile`: Path to modified file
- `changeDescription`: Description of changes
**Output:**
- Affected components list
- Required updates in other documents
- Specific update prompts for each affected document
- Validation checklist
## MCP Integration
All tools are available through MCP when wassden is configured in Claude Code or other MCP clients.
### Tool Names in MCP
| CLI Command | MCP Tool Name |
|------------|--------------|
| `prompt-requirements` | `prompt_requirements` |
| `prompt-design` | `prompt_design` |
| `prompt-tasks` | `prompt_tasks` |
| `prompt-code` | `prompt_code` |
| `generate-review-prompt` | `generate_review_prompt` |
| `validate-requirements` | `validate_requirements` |
| `validate-design` | `validate_design` |
| `validate-tasks` | `validate_tasks` |
| `get-traceability` | `get_traceability` |
| `analyze-changes` | `analyze_changes` |
### Using with Claude Code
Once configured, tools appear in Claude Code's tool palette. Example usage:
```
User: Check if my project description is complete
Assistant: I'll analyze your project description for completeness.
[Uses prompt_requirements tool]
Based on the analysis, your project description is missing:
1. Specific authentication methods
2. Database technology preferences
3. Performance requirements
...
```
## Language Support
### Automatic Detection
All tools automatically detect language from:
1. User input content (for prompts)
2. Document section headers (for validation)
3. Content analysis using pycld2
### Manual Override
Force specific language output:
```bash
# Force Japanese
--language ja
# Force English
--language en
```
### Detection Priority
1. Explicit `--language` parameter (highest)
2. Document content patterns
3. User input analysis
4. Default to Japanese (fallback)
## Performance Characteristics
All tools are optimized for speed:
| Tool | Average Response Time |
|------|---------------------|
| prompt_requirements | <0.003ms |
| prompt_* (others) | <0.003ms |
| validate_* | <0.02ms |
| get_traceability | <0.02ms |
| analyze_changes | <0.02ms |
## Error Handling
### Common Errors
1. **File Not Found**
```
Error: File 'specs/requirements.md' not found
```
Solution: Ensure file exists at specified path
2. **Invalid Format**
```
Error: Invalid EARS format in requirement REQ-01
```
Solution: Follow format guidelines in validation output
3. **Missing Dependencies**
```
Error: Task TASK-01-02 depends on non-existent TASK-01-01
```
Solution: Ensure all referenced IDs exist
### Debug Mode
Enable detailed error output:
```bash
export WASSDEN_DEBUG=1
uv run wassden validate-requirements specs/requirements.md
```
## Best Practices
1. **Start with prompt-requirements**: Ensure project description is complete
2. **Validate after generation**: Always validate agent-generated documents
3. **Maintain traceability**: Use get-traceability regularly
4. **Track changes**: Use analyze-changes when modifying specs
5. **Review implementations**: Use generate-review-prompt for each task
## Examples
### Complete Workflow Example
```bash
# 1. Check project completeness
uvx --from git+https://github.com/tokusumi/wassden-mcp wassden prompt-requirements \
--userInput "Create a REST API for task management with JWT auth"
# 2. Generate requirements (agent uses prompt)
# Agent creates specs/requirements.md
# 3. Validate requirements
uvx --from git+https://github.com/tokusumi/wassden-mcp wassden validate-requirements \
specs/requirements.md
# 4. Generate design prompt
uvx --from git+https://github.com/tokusumi/wassden-mcp wassden prompt-design \
--requirementsPath specs/requirements.md
# 5. Validate design (after agent creates it)
uvx --from git+https://github.com/tokusumi/wassden-mcp wassden validate-design \
specs/design.md
# 6. Generate tasks prompt
uvx --from git+https://github.com/tokusumi/wassden-mcp wassden prompt-tasks \
--designPath specs/design.md
# 7. Validate tasks
uvx --from git+https://github.com/tokusumi/wassden-mcp wassden validate-tasks \
specs/tasks.md
# 8. Check traceability
uvx --from git+https://github.com/tokusumi/wassden-mcp wassden get-traceability
# 9. Review specific task implementation
uvx --from git+https://github.com/tokusumi/wassden-mcp wassden generate-review-prompt TASK-01-01
```
## Related Documentation
- [Specification Format Guide](spec-format.md)
- [Validation Rules](validation/)
- [Development Guide](development.md)
- [Performance Benchmarks](performance.md)
