Engram

Provides contextual analysis of code changes by surfacing historical dependencies and test intents through an MCP server.

typescript

0

GitHub Stars

typescript

Language

4 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "spectra-g-engram": {
      "command": "npx",
      "args": [
        "-y",
        "@spectra-g/engram-adapter"
      ],
      "env": {
        "YOUR_PLACEHOLDER": "<YOUR_PLACEHOLDER>"
      }
    }
  }
}

Engram is an MCP server that provides context to AI agents by surfacing historical dependencies and test intents from your repository. It runs locally, respects privacy, and gives your AI a deeper awareness of how code changes propagate and how tests constrain behavior, so you can reason more effectively about edits.

How to use

You connect Engram to an MCP client to enable enhanced context during coding tasks. Start Engram as an MCP server and configure your client (such as Claude Desktop or Cursor) to communicate over the MCP protocol. When you modify code, Engram can reveal impacted files, associated test intents, and stored memories that guide safe, context-aware changes.

How to install

Prerequisites: make sure you have Node.js and npm installed on your system.

Install Engram’s MCP adapter globally. Run this command in your terminal.

npm install -g @spectra-g/engram-adapter

Configuration and usage patterns

Engram exposes its capabilities to MCP clients through a standard stdio-based command. You configure your MCP client to connect to Engram via a simple command that launches the adapter. Use the example below as a starting point to configure your client.

{
  "mcpServers": {
    "engram": {
      "command": "npx",
      "args": ["-y", "@spectra-g/engram-adapter"]
    }
  }
}

Starting and connecting from clients

To connect from Claude Desktop, add a new MCP server with an MCP URL/command that uses the Engram adapter. For Cursor, configure a new MCP Server with the same stdio command to launch Engram locally.

Additional notes

Engram emphasizes privacy and performs indexing on-demand with low footprint. It includes a knowledge graph for persistent memories and a blast-radius analysis to identify which files are most likely affected by changes.

Available tools

get_impact_analysis

Blast radius calculation for a target file, returning impacted files, test intents, and stored notes.

save_project_note

Remember context about files by storing persistent notes for future impact analyses.

read_project_notes

Retrieve saved context by querying notes related to the project files.