@luminaraptor28/mcp-server

AI agent orchestration for the Model Context Protocol.Built by Lumina28.

LuminaRaptor28 is an AI agent orchestration platform that lets you register agents,compose multi-agent pipelines, query a RAG knowledge base, and trigger n8nworkflow automations — all through a single MCP server that any compatibleclient (Claude Desktop, Cline, Cursor, etc.) can connect to.

Installation

npx (zero-install)

npx @luminaraptor28/mcp-server

Global install

npm install -g @luminaraptor28/mcp-server
luminaraptor28-mcp

From source

git clone https://github.com/lumina28/luminaraptor28-mcp.git
cd luminaraptor28-mcp
npm install
npm run build
npm start

Configuration

The server connects to a LuminaRaptor28 API instance. Configure it with environmentvariables:

Variable	Default	Description
`LUMINARAPTOR28_API_URL`	`http://localhost:1878`	Base URL of the LuminaRaptor28 API
`LUMINARAPTOR28_API_KEY`	(empty)	API key for authentication

Claude Desktop

Add the following to your claude_desktop_config.json:

{
  "mcpServers": {
    "luminaraptor28": {
      "command": "npx",
      "args": ["-y", "@luminaraptor28/mcp-server"],
      "env": {
        "LUMINARAPTOR28_API_URL": "http://localhost:1878",
        "LUMINARAPTOR28_API_KEY": "your-api-key"
      }
    }
  }
}

Cline / Roo Code

Add to your MCP settings:

{
  "mcpServers": {
    "luminaraptor28": {
      "command": "npx",
      "args": ["-y", "@luminaraptor28/mcp-server"],
      "env": {
        "LUMINARAPTOR28_API_URL": "http://localhost:1878",
        "LUMINARAPTOR28_API_KEY": "your-api-key"
      }
    }
  }
}

Available Tools

Agent Management

Tool	Description
`list_agents`	List available AI agents with capabilities and categories
`dispatch_agent`	Send a task to a specific agent for execution
`get_agent_status`	Check trust score, performance stats, and execution history

Orchestration

Tool	Description
`run_pipeline`	Execute a multi-agent pipeline by ID with step-by-step output chaining
`fan_out`	Dispatch a task to multiple agents in parallel with configurable aggregation
`create_pipeline`	Define a new reusable multi-agent pipeline

Knowledge Base (RAG)

Tool	Description
`search_knowledge`	Search the knowledge base for solutions, patterns, and learnings
`add_knowledge`	Add a problem-solution pair to the knowledge base

Workflow Automation

Tool	Description
`trigger_workflow`	Trigger an n8n workflow by ID with optional input data
`list_workflows`	List available n8n workflows with status and execution stats

Usage Examples

Run a multi-agent pipeline

Use run_pipeline with pipeline_id "security-audit" and task
"Audit the authentication module for vulnerabilities"

Fan out to multiple agents

Use fan_out with agent_ids ["code-reviewer", "security-analyst", "performance-auditor"],
task "Review this pull request for issues", and aggregation "merge"

Search knowledge base

Use search_knowledge with query "database connection pooling best practices"
and domain "engineering"

Trigger a workflow

Use trigger_workflow with workflow_id "deploy-staging"
and data { "branch": "main", "environment": "staging" }

API Compatibility

This MCP server is a thin client that delegates to the LuminaRaptor28 REST API. Itrequires a running LuminaRaptor28 instance. The default API endpoint ishttp://localhost:1878. See theLuminaRaptor28 documentation for setupinstructions.

Endpoints Used

Tool	Method	Path
`list_agents`	GET	`/api/v1/agents`
`dispatch_agent`	POST	`/api/v1/agents/dispatch`
`get_agent_status`	GET	`/api/v1/agents/:id/status`
`run_pipeline`	POST	`/api/v1/pipelines/run`
`fan_out`	POST	`/api/v1/orchestration/fan-out`
`create_pipeline`	POST	`/api/v1/pipelines`
`search_knowledge`	POST	`/api/v1/knowledge/search`
`add_knowledge`	POST	`/api/v1/knowledge`
`trigger_workflow`	POST	`/api/v1/workflows/trigger`
`list_workflows`	GET	`/api/v1/workflows`

Development

npm install
npm run dev       # Watch mode (recompiles on change)
npm run build     # One-shot compilation
npm start         # Run the compiled server

License

MIT -- see LICENSE.

@luminaraptor28/mcp-server