home / mcp / syke mcp server

Syke MCP Server

Provides AI-driven code impact analysis by monitoring files, mapping dependencies, and gating builds to prevent cascading breakage.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "khalomsky-syke": {
      "command": "npx",
      "args": [
        "@syke1/mcp-server@latest"
      ],
      "env": {
        "GEMINI_KEY": "your-gemini-key",
        "OPENAI_KEY": "your-openai-key",
        "ANTHROPIC_KEY": "your-anthropic-key",
        "SYKE_AI_PROVIDER": "gemini",
        "SYKE_LICENSE_KEY": "your-key-here"
      }
    }
  }
}

SYKE is an MCP server that provides AI-driven code impact analysis. It watches how your AI tools touch files, builds a dependency graph, detects cascading breakage, and gates builds to prevent damage from spreading. This helps you understand and manage the impact of code changes across your project when using MCP-compatible AI coding agents.

How to use

You use SYKE with your MCP-compatible AI coding client to gain real-time visibility into how changes affect your codebase. On startup, SYKE scans your project and builds a complete dependency graph. When your AI modifies a file, you can run the pre-build gate to verify safety before proceeding. If changes threaten dependent files, SYKE signals a FAIL verdict and blocks the build to prevent cascading failures. Access the live web dashboard at localhost:3333 to explore the dependency graph and risk indicators.

How to install

Prerequisites: ensure you have Node.js installed so you can run npm or npx commands. You will also need access to the MCP clients you plan to connect (Claude Code, Cursor, Windsurf) and the corresponding API keys if you intend to use AI analysis features.

1. Add SYKE to your MCP configuration for each client you use. Use the following snippets in your MCP config files.

{
  "mcpServers": {
    "syke": {
      "command": "npx",
      "args": ["@syke1/mcp-server@latest"],
      "env": {
        "SYKE_LICENSE_KEY": "your-key-here",
        "GEMINI_KEY": "your-gemini-key",
        "OPENAI_KEY": "your-openai-key",
        "ANTHROPIC_KEY": "your-anthropic-key"
      }
    }
  }
}

{
  "mcpServers": {
    "syke": {
      "command": "npx",
      "args": ["@syke1/mcp-server@latest"],
      "env": {
        "SYKE_LICENSE_KEY": "your-key-here",
        "GEMINI_KEY": "your-gemini-key"
      }
    }
  }
}

{
  "mcpServers": {
    "syke": {
      "command": "npx",
      "args": ["@syke1/mcp-server@latest"],
      "env": {
        "SYKE_LICENSE_KEY": "your-key-here",
        "GEMINI_KEY": "your-gemini-key"
      }
    }
  }
}

Additional installation notes

If the toolchain is missing npx, you can provide the full path to the npx executable as the command in your MCP config.

Notes on usage flow

After configuring SYKE in each client, restart your AI agent so SYKE can detect the project language and build its dependency graph. Then you can view the live dependency graph at the dashboard and use tools such as gate_build and analyze_impact to understand and gate changes.

Configuration

SYKE relies on a set of environment variables for licensing and AI provider keys. These keys can be provided per client configuration, or you can rely on a unified set if you choose to centralize them.

A quick reference to the tools

SYKE provides eight MCP tools to inspect and manage code impact. You can run these to inspect dependencies, assess risk, and monitor changes.

Security and keys

Keep your API keys and license keys secure. Use the SYKE_LICENSE_KEY for licensing, and supply provider keys as needed (GEMINI_KEY, OPENAI_KEY, ANTHROPIC_KEY). You can optionally force a provider with SYKE_AI_PROVIDER to ensure consistent analysis across your project.

Troubleshooting

If the dashboard does not open or the dependency graph does not appear, ensure your MCP client configurations point to the correct mcp-server and that the environment keys are valid. Restart the AI agent after updating configurations.

Notes

SYKE auto-selects the AI provider in the order Gemini, OpenAI, then Anthropic unless you explicitly force a provider. The web dashboard visualizes the dependency graph in real time and highlights risk levels.

Available tools

gate_build

Mandatory pre-build check that returns PASS, WARN, or FAIL before any build or deploy.

analyze_impact

Shows direct and transitive dependents when a file is modified.

check_safe

Quick one-line safety verdict: HIGH, MEDIUM, LOW, or NONE risk.

get_dependencies

Lists internal imports (forward dependencies) of a file.

get_hub_files

Ranks files by how many other files depend on them.

refresh_graph

Re-scan all source files and rebuild the dependency graph.

ai_analyze

Pro AI semantic analysis of a file and its dependents.

check_warnings

Real-time monitoring alerts for file changes that may break dependents.