Workflow Discipline Engine

v3.2.0Coding Agentsstable

Server-enforced workflow discipline for AI agents. An MCP server providing persistent work items, dependency graphs, quality gates, and actor attribution. Schemas define what agents must produce — the server blocks the call if they don't. Works with

ai-coding-assistantai-developmentai-harnessai-memoryai-tools
Share:
186
Stars
0
Downloads
0
Weekly
0/5

What is Workflow Discipline Engine?

Workflow Discipline Engine is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to server-enforced workflow discipline for ai agents. an mcp server providing persistent work items, dependency graphs, quality gates, and actor attribution. schemas define what agents must produce — the...

Server-enforced workflow discipline for AI agents. An MCP server providing persistent work items, dependency graphs, quality gates, and actor attribution. Schemas define what agents must produce — the server blocks the call if they don't. Works with

This server falls under the Coding Agents category on MCPgee, the world's largest MCP server directory with 33,000+ servers.

Features

  • Server-enforced workflow discipline for AI agents. An MCP se

Use Cases

Persistent work items and dependencies
Quality gates for agent outputs
Actor attribution and tracking
jpicklyk

Maintainer

LicenseMIT License
Languagekotlin
Versionv3.2.0
UpdatedMay 21, 2026
Statushealthy
Maintenanceactive

Works with

ClaudeOpenAIwindowsmacoslinux

Installation

NPM

npx -y task-orchestrator

Manual Installation

npx -y task-orchestrator

Configuration

Configuration Details

Config File

claude_desktop_config.json

Performance

Response Metrics

Response Time< 200ms
ThroughputMedium

Resource Usage

Memory UsageLow
CPU UsageLow

How to Set Up and Use Workflow Discipline Engine

Task Orchestrator is an MCP server that enforces structured workflow discipline on AI agents by providing persistent work items, dependency graphs, quality gates, and actor attribution backed by a SQLite database. Unlike in-memory task lists, it blocks agent transitions if schema rules or dependency constraints aren't satisfied — ensuring agents produce what they're supposed to before moving on. It runs as a Docker container and integrates with Claude Code, Claude Desktop, and any MCP-compatible client, making it ideal for multi-session and multi-agent development projects where continuity and auditability matter.

Prerequisites

  • Docker installed and running (the server ships as a Docker image, not a native binary)
  • An MCP-compatible client: Claude Code, Claude Desktop, or any client supporting stdio MCP
  • Git (optional, for mounting project-level schema config files)
  • Basic familiarity with Docker volumes for persisting the SQLite database between sessions
1

Pull the Docker image

Download the latest Task Orchestrator image from the GitHub Container Registry. This image contains the Kotlin-based MCP server and its SQLite runtime.

docker pull ghcr.io/jpicklyk/task-orchestrator:latest
2

Register with Claude Code via the CLI

Add the Task Orchestrator to Claude Code's MCP configuration. The -v flag mounts a named Docker volume so the SQLite database persists across container restarts.

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"]
}'
3

Or add to .mcp.json for other clients

For Claude Desktop or other MCP clients, add the server configuration to your .mcp.json or claude_desktop_config.json file.

{
  "mcpServers": {
    "mcp-task-orchestrator": {
      "command": "docker",
      "args": ["run", "--rm", "-i", "-v", "mcp-task-data:/app/data",
        "ghcr.io/jpicklyk/task-orchestrator:latest"]
    }
  }
}
4

(Optional) Enable schema enforcement with a project config

Mount a .taskorchestrator/ directory from your project to activate workflow gates. The AGENT_CONFIG_DIR env var tells the server where to find config.yaml with your custom schemas and gate rules.

{
  "mcpServers": {
    "mcp-task-orchestrator": {
      "command": "docker",
      "args": ["run", "--rm", "-i",
        "-v", "mcp-task-data:/app/data",
        "-v", "/your/project/.taskorchestrator:/project/.taskorchestrator:ro",
        "-e", "AGENT_CONFIG_DIR=/project",
        "ghcr.io/jpicklyk/task-orchestrator:latest"]
    }
  }
}
5

Start using the 14 MCP workflow tools

Once connected, the AI agent can call manage_items, advance_item, manage_dependencies, get_context, get_next_item, claim_item, and more. The server enforces gate rules and dependency ordering automatically.

Workflow Discipline Engine Examples

Client configuration (Claude Desktop / .mcp.json)

Minimal working config using a named Docker volume for persistence. No additional environment variables are required for basic operation.

{
  "mcpServers": {
    "mcp-task-orchestrator": {
      "command": "docker",
      "args": [
        "run", "--rm", "-i",
        "-v", "mcp-task-data:/app/data",
        "ghcr.io/jpicklyk/task-orchestrator:latest"
      ]
    }
  }
}

Prompts to try

Example prompts that exercise Task Orchestrator's workflow and dependency management tools.

- "Get my current work context — what items are active, blocked, or pending?"
- "Create a new work item for implementing user authentication and add it to the backlog"
- "Mark task a3f2 as started and assign it to me"
- "Show all blocked items and explain what they're waiting on"
- "Add a note to item b7c1 documenting the design decision we just made"
- "What's the next available task I should work on?"

Troubleshooting Workflow Discipline Engine

Docker container exits immediately when launched by the MCP client

Ensure the -i flag is present in the docker run args — it keeps stdin open for stdio MCP communication. Without it the container exits right away. Also verify Docker is running before starting your MCP client.

Work items are lost between sessions

Confirm the named volume mcp-task-data is being mounted with -v mcp-task-data:/app/data. Without this volume mount, the container's SQLite database is discarded when --rm removes the container. Run docker volume ls to verify the volume exists.

Schema validation errors when agents try to advance work items

Schema enforcement is active when AGENT_CONFIG_DIR is set and a config.yaml exists in .taskorchestrator/. Review the gate rules in your config.yaml and ensure the agent has populated all required fields before calling advance_item.

Frequently Asked Questions about Workflow Discipline Engine

What is Workflow Discipline Engine?

Workflow Discipline Engine is a Model Context Protocol (MCP) server that server-enforced workflow discipline for ai agents. an mcp server providing persistent work items, dependency graphs, quality gates, and actor attribution. schemas define what agents must produce — the server blocks the call if they don't. works with It connects AI assistants to external tools and data sources through a standardized interface.

How do I install Workflow Discipline Engine?

Install via npm with the command: npx -y task-orchestrator. Then add the server configuration to your AI client's JSON config file (e.g., claude_desktop_config.json or .cursor/mcp.json).

Which AI clients work with Workflow Discipline Engine?

Workflow Discipline Engine works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.

Is Workflow Discipline Engine free to use?

Yes, Workflow Discipline Engine is open source and available under the MIT License license. You can use it freely in both personal and commercial projects.

Browse More Coding Agents MCP Servers

Explore all coding agents servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.

Quick Config Preview

{ "mcpServers": { "task-orchestrator": { "command": "npx", "args": ["-y", "task-orchestrator"] } } }

Add this to your claude_desktop_config.json or .cursor/mcp.json

Read the full setup guide →

Ready to use Workflow Discipline Engine?

Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.

33,000+ ServersFree & Open SourceStep-by-Step Guides