Overview

CodeLogic MCP Server exposes CodeLogic’s code- and data-dependency analysis through the Model Context Protocol (MCP). Integrating this server into MCP-enabled IDEs and agent workflows (Copilot/Claude/Windsurf) lets AI assistants call out to CodeLogic for accurate impact analysis, database relationship insight, and CI/CD scan automation. The result: AI responses grounded in your project’s architecture and dependency surface, reducing risky code changes and improving automated suggestions.

This project is intended for developers who run or integrate MCP servers in their environment. It acts as a bridge between an Astral UV runtime and the CodeLogic backend, offering a small set of tools that agents can invoke to answer questions such as “What will this method change impact?” or “Which database columns does this stored procedure touch?”

GitHub: https://github.com/CodeLogicIncEngineering/codelogic-mcp-server

Features

Exposes CodeLogic analysis via MCP-compatible tools
Method- and database-level impact assessments
CI/CD integration helpers: Docker agents, pipeline templates, build metadata
Small, focused toolset designed for agent workflows in IDEs
Configurable via environment variables for different workspaces and servers

Installation / Configuration

Prerequisites:

Install Astral UV (uv/uvx). See Astral docs: https://docs.astral.sh/uv/getting-started/installation/
A reachable CodeLogic server and valid credentials (username/password) with access to a workspace.

Example: minimal mcp.json for VS Code (using uvx)

{
  "servers": {
    "codelogic-mcp-server": {
      "type": "stdio",
      "command": "uvx",
      "args": ["codelogic-mcp-server@latest"],
      "env": {
        "CODELOGIC_SERVER_HOST": "https://myco.app.codelogic.com",
        "CODELOGIC_USERNAME": "[email protected]",
        "CODELOGIC_PASSWORD": "supersecret",
        "CODELOGIC_WORKSPACE_NAME": "my-workspace",
        "CODELOGIC_DEBUG_MODE": "true"
      }
    }
  }
}

MacOS workaround (uvx issues in some IDEs)

Clone the repo locally.
Point mcp.json to the local uv binary (uv instead of uvx) and run the server from the checkout:

{
  "mcpServers": {
    "codelogic-mcp-server": {
      "type": "stdio",
      "command": "/path/to/uv",
      "args": [
        "--directory",
        "/path/to/codelogic-mcp-server",
        "run",
        "codelogic-mcp-server"
      ],
      "env": {
        "CODELOGIC_SERVER_HOST": "https://myco.app.codelogic.com",
        "CODELOGIC_USERNAME": "alice",
        "CODELOGIC_PASSWORD": "secret",
        "CODELOGIC_WORKSPACE_NAME": "my-workspace",
        "CODELOGIC_DEBUG_MODE": "true"
      }
    }
  }
}

Restart your IDE and refresh MCP servers.

Notes:

On some systems you must use the absolute path to uvx/uv (e.g. /home/user/.local/bin/uvx or C:\Users\user\AppData\Local\astral\uvx.exe).
For non-VS Code IDEs, add a similar server entry to the IDE’s MCP configuration (Claude Desktop, Windsurf, etc.).

Available Tools

The server exposes five MCP tools. Inputs shown are the primary parameters the agent will supply.

Tool name	Purpose	Primary inputs
codelogic-method-impact	Impact analysis for a specific method	class name, method signature
codelogic-database-impact	Maps code ↔ database impact	entity type (table/view/column), entity name
codelogic-docker-agent	Produces Docker agent configs for scans	runtime (.NET/Java/TS/SQL), CI system (GitHub/Jenkins/Azure/GitLab)
codelogic-build-info	Generates build metadata and send/ingest commands	git info, build logs, metadata
codelogic-pipeline-helper	Creates full CI/CD pipeline templates with best practices	CI provider, error/notification preferences

Each tool returns structured JSON and a plain-text summary suitable for agents to incorporate into chat responses or actions.

Use Cases

Pre-change impact checks: In an agent-assisted code edit, call codelogic-method-impact with the class/method being edited to get a list of downstream callers, affected services, and potential regressions. Example: “I want to modify OrderService.CalculateTax(Order order). What will change?”
Database refactor planning: Use codelogic-database-impact to discover all code that reads/writes a column you plan to rename or drop. Example: “If we drop orders.total_discount, which functions and queries break?”
CI/CD automation: Use codelogic-docker-agent or codelogic-pipeline-helper to emit pipeline snippets that run CodeLogic scans, handle scan space, and send results to your build system. Example: automatically add a CodeLogic scan stage to GitHub Actions for every PR.
Build tracing and audit: codelogic-build-info gathers git/build metadata and provides steps to push that data into CodeLogic for traceability.

Tips & Troubleshooting

If tools return authentication errors, verify CODELOGIC_SERVER_HOST and credentials in your server config.
Enable CODELOGIC_DEBUG_MODE=true to get more verbose logs from the MCP server.
If your IDE cannot start uvx, try switching to uv and running from a local clone (see MacOS workaround).
Use the GitHub repo for issues, updates, and contributions: https://github.com/CodeLogicIncEngineering/codelogic-mcp-server

This MCP server is intended to make architectural and dependency data first-class for AI assistants, enabling safer, more informed code and database changes directly from developer workflows.

CodeLogic MCP Server for Code and Data Dependencies