Skip to main content
Glama

Spec MCP Server

A Model Context Protocol server designed to streamline development workflows through AI-assisted analysis, planning, and verification.

Features

  • Tech Steering: Automated codebase analysis and documentation

  • Plan Generation: Comprehensive requirement and design documentation

  • Task Breakdown: Detailed task generation with dependencies

  • Implementation Verification: Automated code review and compliance checking

Related MCP server: Orchestro

Installation

  1. Configure your AI client (Claude Desktop, Cursor, etc.)

  2. Add the server to your MCP configuration

  3. Start using the tools through your AI interface

Configuration Examples

Visual Studio Code

Add to your VS Code MCP configuration file:

  • Regular VS Code: ~/Library/Application Support/Code/User/mcp.json

  • VS Code Insiders: ~/Library/Application Support/Code - Insiders/User/mcp.json

{
  "servers": {
    "spec": {
      "command": "npx",
      "args": [
        "-y",
        "spec-mcp@latest"
      ]
    }
  }
}

Zed

  1. Open Zed > Settings > Open Settings (it will open ~/. config/zed/settings.json)

  2. Add a context_servers section to your configuration:

{
  "context_servers": {
    "spec": {
      "source": "custom",
      "command": "npx",
      "args": ["-y", "spec-mcp@latest"],
      "env": {}
    }
  }
}

Claude Code (CLI)

For Claude Code CLI, use the following command:

claude mcp add spec-mcp --scope user -- npx -y spec-mcp@latest

Claude Desktop

Add to ~/Library/Application Support/Claude/claude_desktop_config.json:

{
  "mcpServers": {
    "spec": {
      "command": "npx",
      "args": [
        "-y",
        "spec-mcp@latest"
      ]
    }
  }
}

Workflow

The Spec MCP workflow guides you through complex development tasks like framework migrations, feature implementation, or refactoring. Follow these steps:

1. Analyze Codebase (One-time Setup)

Analyze the existing codebase to create steering documents that guide all future operations.

Prompt: Use Spec MCP to analyze the codebase

This creates three steering documents in .spec/steering/:

  • product.md - Product overview and business context

  • tech.md - Technology stack and development guidelines

  • structure.md - Project organization and patterns

Note: You can recreate these documents anytime with force_regenerate=true

2. Search Documentation

Search documentation for relevant information about frameworks, APIs, or migration guides using e.g. Context7 MCP.

Prompt: Search docs with Context7 MCP for [topic/framework/library]

3. Generate Tasks

Ask the Spec MCP server to generate a clear, evidence‑driven task list from your goal.

Having your previous analysis in context window you can create tasks from it.

Example prompt:

  • Generate tasks from the analysis using Spec MCP

Output:

  • .spec/specs/tasks.md (includes overview, requirements with EARS, implementable tasks, dependencies, and tests)

4. Implement Tasks

Execute tasks systematically using the task orchestrator, which handles dependencies and parallelization.

Prompt: Implement tasks from tasks.md using Spec MCP task orchestrator

The orchestrator will:

  • Identify ready tasks based on dependencies

  • Execute tasks through task-executor

  • Verify completion with task-checker

  • Report progress and next available tasks

Available Tools

generate-codebase-analysis

Analyzes codebase and generates three foundational analysis documents: product.md, tech.md, and structure.md in .spec/steering/ directory. These documents provide comprehensive analysis of the product features, technology stack, and project structure.

generate-tasks

Spec MCP: Generate tasks.md (Plan & Task Breakdown)

Generate tasks.md (implementation plan & task breakdown) from your request and codebase. Writes .spec/specs/tasks.md with overview, numbered requirements using EARS, implementable tasks (T‑1…), dependencies/phases, and testing guidance. Use for “task breakdown”, “create tasks.md”, “implementation plan”, or “roadmap”. Inputs:

  • user_request (required)

  • project_path (optional, defaults to .)

task-orchestrator

Analyzes tasks.md to identify dependencies, parallelization opportunities, and coordinate task execution. Returns a structured execution plan for deploying task executors efficiently.

task-executor

Executes a specific task from tasks.md by providing detailed implementation guidance, requirements, acceptance criteria, and code patterns. This tool focuses on implementing one task thoroughly.

task-checker

Checks if a task can be marked as done by verifying all acceptance criteria checkboxes are checked and optionally running build/tests. Returns PASS or FAIL with clear reasons.

License

MIT

Available Tools

5 tools
generate-codebase-analysisGenerate Codebase Analysis DocumentsA

Analyzes codebase and generates three foundational analysis documents: product.md, tech.md, and structure.md in .spec/steering/ directory. These documents provide comprehensive analysis of the product features, technology stack, and project structure. Checks if files already exist and skips generation unless force_regenerate is true.

Inputs: additional_context, force_regenerate, project_path.

generate-tasksSpec MCP: Generate tasks.md (Plan & Task Breakdown)A

Generate tasks.md (implementation plan & task breakdown) from your request and codebase. Writes .spec/specs/tasks.md with overview, numbered requirements using EARS, implementable tasks (T-1…), dependencies/phases, and testing guidance. Use for “task breakdown”, “create tasks.md”, “implementation plan”, or “roadmap”.

Inputs: user_request (required), project_path.

task-checkerCheck if Task is CompleteC

Prompt-only verification: instructs the agent to verify acceptance criteria checkboxes for a task with quoted evidence. Returns a PASS/FAIL decision rule to apply.

Inputs: task_id (required), project_path.

task-executorTask ExecutorB

Executes a specific task from tasks.md by providing detailed implementation guidance, requirements, acceptance criteria, and code patterns. This tool focuses on implementing one task thoroughly.

Inputs: task_id (required), project_path, update_status.

task-orchestratorTask OrchestratorA

Use this to implement/execute existing tasks from tasks.md that have already been defined. This tool assumes tasks already exist and focuses on coordinating their implementation. DO NOT use this for generating new plans or tasks - only for executing tasks that have already been created.

Inputs: project_path.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A3.5/5.0

    Scored across 5 tools

    Disambiguation3/5

    The tools have overlapping purposes that could cause confusion, particularly between task-executor and task-orchestrator which both handle task implementation with unclear boundaries. However, descriptions help clarify that task-orchestrator is for coordinating multiple existing tasks while task-executor focuses on single-task implementation, and the other tools have more distinct roles.

    Naming Consistency4/5

    Tool names follow a consistent verb_noun pattern (e.g., generate-codebase-analysis, generate-tasks, task-checker) with clear, descriptive naming. The only minor deviation is that 'task-checker' and 'task-executor' use hyphenated compound nouns rather than separate verb_noun pairs, but the pattern remains readable and predictable throughout.

    Tool Count5/5

    With 5 tools, the count is well-scoped for the server's purpose of specification and task management. Each tool earns its place by covering distinct phases: analysis generation, task planning, verification, and implementation, making the set neither too thin nor overloaded for the domain.

    Completeness4/5

    The tool surface covers the core workflow from codebase analysis to task execution with minor gaps. It includes analysis generation, task planning, verification, and implementation tools, but lacks explicit tools for updating or deleting generated documents or tasks, which agents might need to work around by regenerating files.

    Maintenance

    ActivityInactive
    ResponsivenessNo recent community issues

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      Not graded
      quality
      D
      maintenance
      Enables AI assistants to manage complex development workflows by creating structured handoffs between strategic planning and tactical implementation, with project management, task tracking, and intelligent scaffolding.
      16 npm
      MIT
    • A
      license
      Not graded
      quality
      A
      maintenance
      Enables AI assistants to research, plan, generate requirements and user stories, create starter projects, analyze codebases, curate context, manage tasks, and execute multi-step development workflows.
      69 npm
      MIT