How does Engram differ from Claude Code's built-in memory?

Engram uses a dedicated backend for hybrid search and cross-project accessibility, whereas built-in memory is typically local, project-specific markdown files.

Can I use Engram memory across different repositories?

Yes, memories can be scoped globally, allowing you to share insights, patterns, and configuration settings across your entire local development environment.

Do I need to manually save every observation?

No, Engram uses automatic hooks to capture session summaries and tool usage patterns. You only need to use explicit storage for specific decisions, insights, or credentials.

How do I verify that Engram is connected correctly?

The most reliable method is calling the check_system_health() tool within your Claude Code session to verify the MCP server connection.

Is my data secure when using the Engram vault?

Yes, Engram uses AES-256-GCM encryption for credentials. Values are stored encrypted at rest and are never surfaced in standard search results or context injections.

Engram Memory Management

Name: Engram Memory Management
Author: thebtf

bythebtf

•

開発者ツール

Maintains a persistent, searchable knowledge base across Claude Code sessions to track decisions, patterns, and credentials.

Engram enhances Claude Code by providing a robust, persistent memory layer that goes beyond session-specific context. It automatically captures insights from coding sessions via hooks while offering advanced tools for explicit knowledge storage, semantic retrieval, and encrypted credential management. By integrating hooks that fire during session start, tool use, and termination, it ensures that architectural decisions, recurring patterns, and project history are always accessible, enabling a more coherent and context-aware development experience across multiple projects and repositories.

主な機能

011 GitHub stars

02Cross-project memory sharing to leverage insights globally

03Encrypted AES-256-GCM vault for secure credential and API key storage

04Knowledge curation tools for merging duplicates and maintaining data quality

05Hybrid semantic and full-text search for precise knowledge retrieval

06Automatic observation capture via SessionStart and Stop hooks

ユースケース

01Quickly onboarding onto unfamiliar files by reviewing past observations and explanations

02Tracking architectural decisions and 'why' behind specific implementation choices

03Managing and retrieving API tokens securely without hardcoding or environment leaks

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add thebtf/engram memory

For use in Claude.ai and ChatGPT