MCP Server & Claude Code Plugin for Multi-Agent Workflows
An MCP server that orchestrates AI agents through structured markdown documents. Create linked specifications and tasks, assign work to specialized agents with a single command, and let the system handle context injection and workflow management automatically.
This server enables you to:
- Spec out projects as a team or solo using interlinked markdown documents
- Make better decisions using the multi-option tradeoff workflow
- Orchestrate specialized agents by simply assigning tasks—the server injects the right context automatically
- Maintain impartial reviews by keeping the coordinator agent separate from implementation details
- Create linked documents with specifications, guides, and architecture decisions
- Add @references to link related content that gets auto-injected when needed
- Assign tasks to agents using coordinator or subagent workflows
- Agent completes and reports "Done"—coordinator reviews code changes directly
- Review without bias by examining the actual changes, only consulting notes if needed
The system preserves context across sessions while keeping your main agent focused on orchestration and review rather than implementation details.
Dual Task System
- Coordinator tasks for sequential project work (auto-archives when complete)
- Subagent tasks for flexible ad-hoc work across documents
Automatic Context Injection
- Link documents with @/path/doc.md#sectionsyntax
- Referenced content loads automatically when tasks start
- Works on any project—no configuration needed
Workflow Library
- 9 pre-built workflows for common development scenarios
- Access via get_workflowtool or Claude Code plugin commands
- Reference workflows in task metadata for automatic injection
- Create your own custom workflows easily
- Node.js 18+
- pnpm 10.x
pnpm install
pnpm build# Development
pnpm dev
# Production
pnpm start
# Test with inspector
pnpm inspectorSet environment variables in your MCP client config:
{
  "mcpServers": {
    "ai-prompt-guide-mcp": {
      "command": "npx",
      "args": ["-y", "ai-prompt-guide-mcp"],
      "env": {
        "MCP_WORKSPACE_PATH": "./.ai-prompt-guide"
      }
    }
  }
}Required:
- MCP_WORKSPACE_PATH- Path to your project workspace (absolute or relative to project root)
Optional - Project-Specific Folders:
These folders store your project data and default to subdirectories within MCP_WORKSPACE_PATH:
- DOCS_BASE_PATH- Documents and subagent tasks (default:- {workspace}/docs)
- ARCHIVED_BASE_PATH- Completed work (default:- {workspace}/archived)
- COORDINATOR_BASE_PATH- Sequential tasks (default:- {workspace}/coordinator)
Optional - Plugin-Global Resources:
These folders contain reusable workflows and guides, independent of MCP_WORKSPACE_PATH:
- WORKFLOWS_BASE_PATH- Workflow protocols (default: bundled with plugin)
- GUIDES_BASE_PATH- Documentation guides (default: bundled with plugin)
Optional - Other Settings:
- REFERENCE_EXTRACTION_DEPTH- How deep to follow @references (1-5, default 3)
- LOG_LEVEL- Logging verbosity (debug, info, warn, error)
Path Resolution:
- Relative paths are resolved from MCP_WORKSPACE_PATH(for project folders) or plugin directory (for workflows/guides)
- Absolute paths are used as-is
- Example: DOCS_BASE_PATH: "custom-docs"→{workspace}/custom-docs
- Example: DOCS_BASE_PATH: "/abs/path/docs"→/abs/path/docs
Per-Project Configuration:
Use .mcp-config.json in your project root to override paths without modifying MCP client settings:
{
  "env": {
    "MCP_WORKSPACE_PATH": "./.ai-prompt-guide",
    "DOCS_BASE_PATH": "documentation",
    "ARCHIVED_BASE_PATH": "archive"
  }
}This is useful when different projects need different folder structures.
Default structure (created automatically):
.ai-prompt-guide/
├── docs/               # Your documents and subagent tasks
├── coordinator/        # Sequential project tasks
├── archived/          # Completed work (auto-populated)
│   ├── docs/          # Archived documents
│   └── coordinator/   # Archived task lists
Plugin-global resources (bundled, rarely customized):
{plugin-dir}/.ai-prompt-guide/
├── workflows/         # Reusable workflow protocols
└── guides/           # Documentation best practices
Only docs/ is required in your project—everything else is created automatically. Workflows and guides are provided by the plugin unless you override them.
/plugin marketplace add https://github.com/Blakeem/AI-Prompt-Guide-MCP
/plugin install ai-prompt-guideThe plugin provides 9 workflows accessible both as slash commands and via the get_workflow MCP tool:
- /ai-prompt-guide:develop-tdd– Orchestrate multi-agent development with TDD
- /ai-prompt-guide:develop-iterate– Orchestrate multi-agent development with manual verification
- /ai-prompt-guide:review– Targeted review of PRs or components
- /ai-prompt-guide:audit– Quality audit with specialized agents
- /ai-prompt-guide:coverage– Add comprehensive test coverage
- /ai-prompt-guide:decide– Structured decision making with trade-off analysis
- /ai-prompt-guide:decide-iterate– Multi-perspective decision analysis with parallel agents
- /ai-prompt-guide:spec-feature– Document internal feature specifications
- /ai-prompt-guide:spec-external– Document external API specifications
Commands are shortcuts to workflows. When using Claude Code, the plugin commands provide a convenient way to invoke workflows. When using the MCP server directly, access the same workflows via:
get_workflow({ workflow: "develop-tdd" })
get_workflow({ workflow: "audit" })
// ... etc/ai-prompt-guide:develop-tdd Build an admin dashboard with user activity charts, region filtering, and CSV export. Include tests for the aggregation logic.
The plugin loads the workflow, creates a plan, assigns work to specialized agents, and orchestrates implementation automatically.
The server provides 20 MCP tools organized by function:
- create_document– Create new documents with namespace selection
- browse_documents– Navigate document hierarchy and list contents
- search_documents– Full-text or regex search across all documents
- section– Edit, append, insert, or remove sections in bulk
- coordinator_task– Create, edit, or list coordinator tasks
- start_coordinator_task– Start the first pending task with full context
- complete_coordinator_task– Complete task and get next or auto-archive
- view_coordinator_task– View coordinator task details
- subagent_task– Create, edit, or list subagent tasks
- start_subagent_task– Start specific task with full context
- complete_subagent_task– Complete task and get next pending
- view_subagent_task– View subagent task details
- view_document– View complete document structure with metadata
- view_section– View section content without starting work
- edit_document– Update document title and overview
- delete_document– Delete or archive documents
- move– Move sections within or across documents (supports both regular sections and subagent tasks)
- move_document– Move documents to new namespaces
- get_workflow– Load workflow protocol content
- get_guide– Access documentation guides
All tools use consistent addressing (/doc.md#section) and work together seamlessly.
Spec-driven development with automatic agent orchestration:
- Create linked specification documents for your project
- Add coordinator tasks for the implementation phases
- Assign the first task to a specialized agent with one command
- Agent gets full context automatically (specs, workflows, references)
- Agent completes work and reports "Done"
- Review actual code changes to maintain objectivity
- Move to next task—system queues it with the right context
The coordinator agent stays focused on orchestration and quality while specialized agents handle implementation. Your impartiality is preserved because you review code directly, not summaries.
MIT. See LICENSE for details.