NickCirv

NickCirv / engram

Public

The context spine for AI coding agents. 6 providers (graph + git + MemPalace + Context7 + Obsidian) assembled into rich packets per Read. Up to 90% session-level token savings. Local SQLite, zero native deps, zero cloud.

www.npmjs.compackageengramx agent-tools ai claude-code claude-code-hooks code-intelligence
11
2
100% credibility
Found Apr 13, 2026 at 11 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
TypeScript
AI Summary

engram builds a local knowledge graph from your codebase and intercepts AI coding tool calls to deliver rich structural context, reducing token usage by up to 90%.

How It Works

1
🔍 Discover engram

You hear about a helpful tool that remembers your code structure so your AI coding buddy gets smarter without rereading everything.

2
📦 Get it set up

With one easy step, you add engram to your computer so it's ready wherever you work.

3
📁 Prepare your project

In your coding folder, you tell engram to look at your files and build a smart map of how everything connects.

4
🔌 Link to your AI helper

You connect engram to your AI coding tool with a quick setup, unlocking automatic smart help.

5
💬 Start chatting with AI

Open your AI coding session in the project, and it automatically pulls in key info without you asking.

6
⚡ Get instant insights

Instead of loading whole files, your AI sees quick summaries of structure, past issues, and changes, speeding everything up.

🎉 Code smarter, faster

Your AI now remembers your project's layout perfectly, saving time and making fixes and features easier every session.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 11 to 11 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is engram?

Engram is the context spine for AI coding agents, building a local SQLite knowledge graph from codebase structure, git history, decisions, and mistakes across 10 languages. Written in TypeScript with zero native deps or cloud, it scans repos via `engram init` and intercepts Claude Code tool calls like Read, serving rich packets—structure, docs, issues—in one ~500-token response. Users get up to 90% session token savings without agents needing extra tools.

Why is it gaining traction?

It flips passive Reads into automatic graph summaries via hooks, closing the gap where agents forget to query context github copilot or workflows. Benchmarks pit it against MEMORY.md and Cursor memory, showing 3-11x savings vs relevant files. Git hooks keep graphs fresh on commits, and MCP server exposes queries for any client, plus integrations like Context7 and Obsidian for engram memory github depth.

Who should use this?

Claude Code users tired of token burn on repeated file reads in long sessions. Agent builders needing github context engineering hub for structure-aware prompts. Teams tracking engrams of past mistakes before refactors or features.

Verdict

Promising early project at 11 stars and 1.0% credibility—docs shine with diagrams and tiers, 520 tests pass, but low adoption signals maturity risks. Try on a side project if context github actions matter; uninstall is clean.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.