# Graphify [Health: Active]

**Category:** 🧠 Knowledge & Memory  
**Repository:** https://github.com/Graphify-Labs/graphify-mcp  
**GitHub Stars:** 0  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/graphify

## Description
Search indexed code, trace dependencies, assess change impact, and recall repository memory.

## Claude Desktop Quick Installation
No confirmed install details yet, so no config is published. Follow the setup steps in the repository above.

## Documentation

## What Graphify MCP server does

Graphify provides an MCP connection to a hosted code knowledge graph. It lets an AI assistant discover available workspaces and repositories, locate implementations, inspect relationships around symbols, estimate what a code change may affect, identify linked tests, and store or retrieve repository decisions.

The Graphify MCP server is intended for repositories connected to a Graphify workspace and indexed before use. It can support investigations such as finding callers of a function, understanding dependency paths, or checking whether relevant tests are associated with a symbol. Memory features can preserve repository knowledge between sessions, including decisions that an assistant should recall later.

## How it works

The service is reached remotely through the Streamable HTTP endpoint `https://api.graphify.com/mcp`. Authentication uses Graphify OAuth in the browser rather than a password or static token in the client configuration. The client sends MCP requests to the hosted service after authorization, and Graphify returns results from the repositories and index data available to the authenticated workspace.

Before querying code, sign in at Graphify, choose a workspace, connect a repository you are allowed to use, and wait for indexing to finish. A successful protocol connection alone does not confirm repository access or validate the accuracy of an individual result.

## Setup and configuration

To connect the Graphify MCP server in VS Code, use the Command Palette’s **MCP: Add Server** command, select **HTTP**, and enter the service URL. You can also place this configuration in `.vscode/mcp.json`:

```json
{
  "servers": {
    "graphify": {
      "type": "http",
      "url": "https://api.graphify.com/mcp"
    }
  }
}
```

Let the client complete the browser OAuth flow and manage credentials. Do not put a password or access token in the configuration file. Other clients with remote MCP forms can use the same URL, though configuration details vary. Clients that only accept STDIO may use `mcp-remote@0.14.3` through `npx`; that bridge requires Node.js and npm and stores OAuth credentials locally.

## Tools and capabilities

The Graphify MCP server supports workflows for:

- Listing Graphify workspaces and repositories.
- Finding a function or other implementation in a selected repository.
- Following callers, dependencies, and other code relationships.
- Estimating static change impact and locating linked tests.
- Remembering repository decisions and recalling them later.

These capabilities are useful for code navigation, dependency investigation, change planning, and maintaining context across repository work.

## Limitations and notes

Static impact analysis and linked-test results guide investigation; they do not run tests or establish runtime correctness. Memory requests may create or modify repository knowledge, so review write operations before approving them in the client.

The Graphify MCP server only exposes information available through the user’s permissions, selected workspace, connected repositories, and completed index. Empty results can indicate missing access or incomplete indexing. Model access and tool enablement are controlled separately by the AI client, while Graphify service plans apply independently. Do not publish credentials, tokens, private repository contents, or sensitive logs in public issues.

_Full upstream README: https://allmcps.com/mcp/graphify/readme_