task-orchestrator

Overview

Task Orchestrator is an MCP server that implements server-enforced workflow discipline for multi-agent AI systems. It provides a persistent work item graph with quality gates that prevent progression until requirements are met. Unlike prompt-based frameworks that rely on models following instructions, this server blocks API calls when dependencies aren't satisfied or required documentation is missing. It features actor attribution for accountability, dependency graph validation, and session continuity through a SQLite database backend.

Try asking AI

After installing, here are 7 things you can ask your AI assistant:

• manage_items

  Create, update, or delete work items in the task hierarchy

  Try:Create a new work item for implementing user authenticationTry:Update the title of the work item with ref 'auth'Try:Delete the completed work item 'a3f2'
• query_items

  Search for and retrieve work items based on various criteria

  Try:Show me all work items in the 'work' phaseTry:Find all items that are currently blockedTry:Get all work items created after January 1st, 2025
• create_work_tree

  Create a hierarchical tree of work items with dependencies

  Try:Create a work tree for the User Authentication feature with child items for schema, API, and testsTry:Build a multi-level breakdown of the payment processing workflowTry:Create a parallel task structure for testing and implementation
• complete_tree

  Mark a work item and all its children as completed

  Try:Complete the entire 'User Authentication' work treeTry:Mark the parent project and all subtasks as finishedTry:Complete all items in the current sprint
• manage_notes

  Create, update, or delete phase-specific documentation attached to work items

  Try:Add requirements notes to the login API taskTry:Update the implementation notes with what was actually builtTry:Delete outdated requirements from the stalled feature task
• query_notes

  Retrieve notes from work items, with optional filtering by role or key

  Try:Show me all the work notes for the 'api' taskTry:Get the review notes for the security assessmentTry:List all requirements notes in the project
• manage_dependencies

  Create or remove dependency relationships between work items

  Try:Create a dependency that the tests task must wait for the API implementationTry:Remove the dependency between the frontend and backend tasksTry:Set up a fan-in dependency where multiple tasks must complete before the next phase
• query_dependencies

  Retrieve dependency relationships between work items

  Try:Show me what tasks are blocking the payment integrationTry:Get all dependencies for the 'api' taskTry:Find which tasks depend on the database schema completion
• advance_item

  Move a work item to the next phase in its lifecycle, enforcing quality gates

  Try:Advance the user authentication task from 'queue' to 'work' phaseTry:Move the API implementation to 'review' phaseTry:Transition the completed task to 'terminal' status
• get_next_status

  Determine the next possible status for a work item based on current state and dependencies

  Try:What's the next possible status for the stalled authentication task?Try:Can the API task advance to work now?Try:Show me available transitions for the blocked feature task
• get_context

  Retrieve the current state of the work item graph with recent changes and actor attribution

  Try:Get all context since yesterday's sessionTry:Show me the current state of the entire projectTry:Get context for the 'api' task with all ancestors
• get_next_item

  Find the next available work item to work on based on dependencies and availability

  Try:Find the next available task for me to work onTry:Get a ready task for the implementation phaseTry:Show me items that can advance from queue to work

Comparable tools

Installation

# Pull the latest image
docker pull ghcr.io/jpicklyk/task-orchestrator:latest

# Register with Claude Code
claude mcp add-json mcp-task-orchestrator '{
  "command": "docker",
  "args": [
    "run", "--rm", "-i",
    "-v", "mcp-task-data:/app/data",
    "ghcr.io/jpicklyk/task-orchestrator:latest"
  ]
}'

# For MCP clients: add to .mcp.json
{
  "mcpServers": {
    "mcp-task-orchestrator": {
      "command": "docker",
      "args": [
        "run", "--rm", "-i",
        "-v", "mcp-task-data:/app/data",
        "ghcr.io/jpicklyk/task-orchestrator:latest"
      ]
    }
  }
}

FAQ

How is this different from prompt-based workflow frameworks?

Unlike prompt frameworks that rely on models following instructions, Task Orchestrator enforces rules at the server level by blocking API calls when requirements aren't met. Rules live in the server, not in prompts.

Can I use this with any AI client?

Yes, it works with any MCP-compatible client. The provided Claude Code plugin adds additional automation features but isn't required for the core functionality.

Does this require a specific development methodology?

No, Task Orchestrator enforces workflow structure without imposing methodology. You define your workflow rules through YAML schemas without prescribing development processes.

Compare task-orchestrator with