@summoniq/mcp-server
v0.1.0
Published
MCP server for SummonIQ Orchestrator - enables AI assistant to interact with projects, configs, and workflows
Readme
SummonIQ Orchestrator MCP Server
Model Context Protocol server that exposes SummonIQ Orchestrator functionality to AI assistants.
Features
This MCP server provides AI assistants with access to:
Project Management
list_projects- List all projects with filtersget_project- Get detailed project informationcreate_project- Create new projects
Team Management
list_teams- List all teams with workflows and agentsget_team- Get detailed team information
Workflow Management
list_workflows- List available workflowsget_workflow- Get detailed workflow information with stages
Agent Management
list_agents- List agents with role filtering- Get agent assignments and capabilities
Task/Ticket Management
list_tickets- List tickets with project/status filtersget_ticket- Get detailed ticket information with requirements
Configuration Management
list_configs- List environment and dependency configurations
Setup
1. Install Dependencies
cd packages/mcp-server
bun install2. Build
bun run build3. Configure Prisma
The MCP server uses the same Prisma client as the orchestrator. Make sure the orchestrator's Prisma is set up:
cd ../../orchestrator
bunx prisma generate4. Add to MCP Settings
Add to your MCP client configuration (e.g., Claude Desktop ~/Library/Application Support/Claude/claude_desktop_config.json):
{
"mcpServers": {
"applab-orchestrator": {
"command": "node",
"args": [
"/path/to/applab/packages/mcp-server/dist/index.js"
],
"env": {
"DATABASE_URL": "file:../../orchestrator/prisma/dev.db"
}
}
}
}Or use with other MCP clients by running:
node dist/index.jsUsage Examples
Once connected, the AI assistant can:
# List all active projects
Use list_projects to show me all active projects
# Get project details
Get details for project "my-web-app" using get_project
# Create a new project
Create a new web-app project called "dashboard" using create_project
# List teams and their workflows
Show me all teams using list_teams
# View available workflows
List all default workflows with list_workflows
# Find tickets
Show me all tickets in "my-web-app" project that are in progressDevelopment
# Watch mode for development
bun run dev
# Build for production
bun run build
# Start the server
bun run startArchitecture
- Transport: stdio (standard input/output)
- Database: SQLite via Prisma (shared with orchestrator)
- Protocol: Model Context Protocol (MCP) v1.0
Tool Capabilities
All tools return JSON responses with full relationship data. Tools validate inputs using Zod schemas and provide helpful error messages.
Response Format
All successful tool calls return:
{
"content": [
{
"type": "text",
"text": "<JSON data>"
}
]
}Error Handling
- Zod validation errors show field-specific error messages
- Database errors indicate missing records or constraint violations
- All errors include descriptive messages for debugging
Integration with AI Assistant Chat
This MCP server is designed to work with the AI assistant chat in the bottom right of the SummonIQ Orchestrator UI. The assistant can:
- Query project status - Check project progress, tickets, and team assignments
- Manage workflows - View and understand workflow stages
- Monitor agents - See which agents are assigned to which tasks
- Track tickets - Find tickets by status, project, or priority
- Analyze configurations - Review environment and dependency setups
The assistant has full read access and limited write access (can create projects, but not delete).
