Corbell Spec Generator

v1.0.0Coding Agentsstable

AI-powered spec generation and review using multi-repo code graph intelligence for backend teams that ship to production.

code-analysiscode-embeddingcode-intelligence-mcpknowledge-graphmcp-server
Share:
210
Stars
0
Downloads
0
Weekly
0/5

What is Corbell Spec Generator?

Corbell Spec Generator is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to ai-powered spec generation and review using multi-repo code graph intelligence for backend teams that ship to production.

AI-powered spec generation and review using multi-repo code graph intelligence for backend teams that ship to production.

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

Features

  • AI-powered spec generation and review using multi-repo code

Use Cases

Generate AI specs from multi-repo code analysis.
Review specifications with code graph intelligence.
Support backend teams with spec automation.
Corbell-AI

Maintainer

LicenseApache-2.0
Languagepython
Versionv1.0.0
UpdatedMay 19, 2026
Statushealthy
Maintenanceactive

Works with

ClaudeOpenAIwindowsmacoslinux

Installation

Manual Installation

npx corbell

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 Corbell Spec Generator

Corbell is an AI-powered spec generation and review tool that builds a multi-repository code graph of your backend codebase and uses it to produce accurate, context-aware technical specifications. It uses tree-sitter for static analysis and sentence-transformers for semantic code search to understand service dependencies, call paths, and typed method signatures across your entire codebase — not just a single file. Backend teams use Corbell to automate the grunt work of writing design docs, review existing specs against what the code actually does, and export finalized specs directly to Jira, Linear, or Notion. It exposes its graph queries and code search capabilities as MCP tools for use inside Cursor, Claude Desktop, and other IDE integrations.

Prerequisites

  • Python 3.11 or higher
  • At least one AI provider API key: ANTHROPIC_API_KEY for Claude or OPENAI_API_KEY for GPT-4o
  • Git repositories containing the backend code you want to analyze
  • An MCP-compatible client such as Claude Desktop or Cursor
  • Optional: Jira API token (CORBELL_JIRA_API_TOKEN), Linear API key (CORBELL_LINEAR_API_KEY), or Notion credentials for spec export
1

Install Corbell with your preferred AI provider

Install Corbell via pip. Include extras for your chosen LLM provider and any export integrations you need.

pip install "corbell[anthropic]"  # for Claude
# or
pip install "corbell[openai]"     # for GPT-4o
# or all integrations:
pip install "corbell[anthropic,openai,notion,linear,jira]"
2

Set environment variables

Export your AI provider key and any integration keys you need. CORBELL_WORKSPACE can point to a parent directory containing multiple repos.

export ANTHROPIC_API_KEY=your_anthropic_key
# or
export OPENAI_API_KEY=your_openai_key

# Optional integrations:
export CORBELL_JIRA_API_TOKEN=your_jira_token
export CORBELL_LINEAR_API_KEY=your_linear_key
export CORBELL_LINEAR_TEAM_ID=your_team_id
export CORBELL_WORKSPACE=/path/to/your/repos
3

Initialize Corbell in your workspace

Run corbell init inside your repository or workspace directory. This sets up the configuration and prepares for graph building.

cd /path/to/your/workspace
corbell init
4

Build the code graph

Build the multi-repo code graph with method-level analysis. This step uses tree-sitter to parse your code and build the knowledge graph that powers spec generation.

corbell graph build --methods
5

Start the MCP server for IDE integration

Launch Corbell's MCP server to expose graph queries and code search to Claude Desktop or Cursor.

corbell mcp serve
6

Configure your MCP client

Add Corbell to your claude_desktop_config.json or Cursor MCP settings to enable AI-assisted spec generation directly in your editor.

{
  "mcpServers": {
    "corbell": {
      "command": "corbell",
      "args": ["mcp", "serve"],
      "env": {
        "ANTHROPIC_API_KEY": "your_anthropic_key",
        "CORBELL_WORKSPACE": "/path/to/your/repos"
      }
    }
  }
}
7

Generate your first spec

Use the CLI to generate a spec for a new feature, pointing Corbell to a PRD file or describing the feature directly.

corbell spec new --feature "Payment Retry Logic" --prd-file prd.md
corbell spec review specs/payment-retry.md

Corbell Spec Generator Examples

Client configuration

Add Corbell to your claude_desktop_config.json. The mcp serve subcommand exposes graph queries, code search, and spec tools over the MCP protocol.

{
  "mcpServers": {
    "corbell": {
      "command": "corbell",
      "args": ["mcp", "serve"],
      "env": {
        "ANTHROPIC_API_KEY": "your_anthropic_key",
        "CORBELL_WORKSPACE": "/path/to/your/backend/repos"
      }
    }
  }
}

Prompts to try

These prompts use Corbell's code graph intelligence to generate specs, analyze dependencies, and review existing design documents.

- "Generate a technical spec for adding a payment retry mechanism to our checkout service"
- "Show me the call graph for the UserAuthService and all services it depends on"
- "Review the spec in specs/payment-retry.md and check if it matches what the code actually does"
- "Find all methods in the codebase that call the sendEmail function"
- "Export the payment retry spec to a Linear issue in our backend team"

Troubleshooting Corbell Spec Generator

corbell graph build fails with parse errors

Corbell uses tree-sitter for static analysis, which requires supported language grammars. Ensure your code is in a supported language (Python, TypeScript, Go, etc.). Check that CORBELL_WORKSPACE is set to the correct directory containing your source repositories.

Spec generation produces generic output not grounded in the codebase

Run corbell graph build --methods before generating specs. Without the method-level graph, Corbell cannot discover service dependencies or typed signatures. Also confirm CORBELL_WORKSPACE points to the repos containing the relevant backend code.

Export to Linear or Jira fails

Verify CORBELL_LINEAR_API_KEY and CORBELL_LINEAR_TEAM_ID (for Linear) or CORBELL_JIRA_API_TOKEN (for Jira) are set correctly. Install the appropriate extras: pip install 'corbell[linear]' or pip install 'corbell[jira]'.

Frequently Asked Questions about Corbell Spec Generator

What is Corbell Spec Generator?

Corbell Spec Generator is a Model Context Protocol (MCP) server that ai-powered spec generation and review using multi-repo code graph intelligence for backend teams that ship to production. It connects AI assistants to external tools and data sources through a standardized interface.

How do I install Corbell Spec Generator?

Follow the installation instructions on the Corbell Spec Generator GitHub repository. Clone the repo, install dependencies, and add the server config to your AI client.

Which AI clients work with Corbell Spec Generator?

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

Is Corbell Spec Generator free to use?

Yes, Corbell Spec Generator is open source and available under the Apache-2.0 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": { "corbell": { "command": "npx", "args": ["-y", "corbell"] } } }

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

Read the full setup guide →

Ready to use Corbell Spec Generator?

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