Skip to main content

​
What is the Web Server?

The sudocode web server is a local web application that provides a visual interface and control plane for managing your sudocode project. It gives you a browser-based view of your specs, issues, and relationships while also serving as a coordination layer for dispatching work to AI agents.
The web server runs locally on your machine (typically at http://localhost:3000) and provides real-time visualization of your project context alongside agent orchestration capabilities.

​
Why Use the Web Server?

Visual Context Management

See your entire project at a glance
  • Rendered markdown specs with live relationships
  • Kanban board for issue tracking
  • Real-time feedback display

Agent Orchestration

Coordinate AI agents like a team
  • Dispatch issues to agents on worktrees
  • Monitor agent progress in real-time
  • Review completed work and feedback

Enhanced Workflows

Beyond what MCP alone provides
  • Visual spec writing and editing
  • Drag-and-drop issue management
  • Interactive dependency management

Local & Private

Your data stays on your machine
  • Runs entirely locally
  • Git-based persistence
  • Connects to your configured coding agent

​
Web Server vs. MCP Server

The web server and MCP server complement each other:
FeatureMCP ServerWeb Server
Primary UseDirect agent integrationVisual management + orchestration
InterfaceAgent chat sessionsBrowser-based UI for humans
Agent AccessBuilt into agent workflowsDispatch agents from UI
VisualizationPlain Markdown files and CLI commandsRich visual interface
Best ForSynchronous agent autonomyHuman oversight, multi-agent coordination
SetupMCP configuration in agent or pluginSimple sudocode server command
The sudocode server uses the MCP server under the hood to enable agent access to the sudocode database. Think of the web server as a control plane and visualization layer on top of MCP functionality.

​
Key Features

​
1. Spec Visualization

View specifications as beautifully rendered markdown with live relationship tracking:
sudocode spec board
  • Click on [[spec-id]] references to navigate
  • See incoming/outgoing relationships
  • View anchored feedback inline
  • Edit specs with live preview

​
2. Issue Management

Kanban-style board for organizing work:
sudocode issues board
  • Drag issues between open/in_progress/blocked/closed
  • Filter by priority, tags, or assignee
  • Dispatch issues to agents
  • Monitor agent progress

​
3. Agent Dispatch

Coordinate multiple AI agents from one interface:
sudocode issues board dispatching
  • Select ready issues and assign to agents
  • Launch agents with specific context
  • Track multiple agents working in parallel
  • Review agent-provided feedback

​
4. Real-Time Updates

The web server uses WebSockets for live updates:
  • See changes as agents work
  • File watcher detects external edits
  • Automatic sync across all connected clients
  • No manual refresh needed

​
How It Works

The flow:
  1. You run sudocode server in your project
  2. Web server starts
  3. Open your browser to http://localhost:3000
  4. You view specs, manage issues, and dispatch agents
  5. Agents connect via MCP and work on issues
  6. Web server displays real-time progress and results

​
Use Cases

Scenario: You’re using Claude Code but want to see your full project context at a glance.Solution: Run the web server alongside Claude Code:
  • Claude Code accesses context via MCP
  • You visualize and organize context in the browser
  • Both stay in sync automatically
Scenario: Your team needs to review specs and coordinate agent work.Solution: Use the web server as a shared view:
  • Team members view specs and issues in browser
  • Agents work through MCP integration
  • All changes sync via git
  • Web server provides real-time updates
Scenario: You want to run multiple agents on different tasks simultaneously.Solution: Dispatch and monitor agents from the web UI:
  • Select ready issues for each agent
  • Launch agents from the interface
  • Monitor progress in real-time
  • Review completed work and feedback
Scenario: You want a better experience for writing and organizing specs.Solution: Use the web server’s spec editor:
  • Visual markdown editor with live preview
  • Auto-complete for [[references]]
  • Drag-and-drop file uploads
  • Instant feedback from agent reviews

​
What’s Included

The web server provides: REST API:
  • CRUD operations for specs, issues, relationships, feedback
  • Query endpoints for ready work, blocked issues, stats
  • Agent execution endpoints for dispatching work
WebSocket Server:
  • Real-time updates for connected clients
  • Live sync with file system changes
  • Agent progress notifications
Web UI:
  • Spec browser with rendered markdown
  • Kanban board for issues
  • Agent dispatch interface
  • Dependency graph visualization
Built-In Agent:
  • Agentic spec writing from the UI
  • Issue creation workflows
  • Automated feedback generation
The web server’s agent capabilities depend on the MCP server being configured. The web server provides the UI and coordination, while agents connect via MCP to access the sudocode database.

Getting Started

Set up and launch your first web server

MCP vs Web Server

Detailed comparison and when to use each

Spec Writing

Write specs using the web interface

Issue Management

Organize and dispatch issues

Feedback Workflows

Review and manage agent feedback

MCP Server Setup

Configure MCP for agent access

​
Next Steps

1

Quick Start

Follow the Quick Start guide to launch your web server in minutes
2

Configure MCP agent

Set up an MCP agent for AI-assisted spec creation and issue management
3

Create your first spec

Use the Spec Writing guide to capture requirements
4

Dispatch your first issue

Learn Issue Management to coordinate agent work