KumihoIO

KumihoIO / construct-os

Public

Memory-native AI agent runtime: Rust gateway + React dashboard + Python Operator + Kumiho graph memory

10
0
100% credibility
Found Apr 28, 2026 at 10 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Rust
AI Summary

Construct is a Rust-native open-source AI agent runtime featuring persistent graph-based memory, multi-agent YAML workflow orchestration, a React web dashboard, hardware peripheral support, and integration with numerous LLM providers.

How It Works

1
📥 Get the easy starter kit

Download the simple installer script and run it to set up your personal AI helper on your computer.

2
🧠 Connect your thinking partner

Pick an AI service like a smart chat helper and add your private access code so it can understand and respond.

3
🌐 Open the colorful dashboard

Launch the web page that shows everything your AI is doing, like a window into its world.

4
🤖 Create your first smart helper

Build a custom AI teammate by describing its personality, skills, and what it's great at.

5
Choose your adventure
💬
Quick chat

Ask your helper a question and watch it think step by step.

🔄
Team workflow

Set up a chain of helpers that pass work to each other like a relay race.

6
📈 Watch the magic happen live

See your helpers light up on the screen as they work, remember everything, and build trust over time.

🎉 Your AI team is alive and learning

Now you have smart helpers that remember past work, get better with use, and handle tasks on their own.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 10 to 10 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 construct-os?

Construct-os is a memory-native AI agent runtime built in Rust, delivering a gateway for agent execution, React dashboard for live monitoring, Python operator for multi-agent orchestration, and Kumiho graph memory for persistent state. Developers get declarative YAML workflows with DAG views, trust-scored agents, session auditing, and hardware peripherals like ESP32 or Arduino—all queryable via graph edges. To construct agents, you define templates, run CLI commands like `construct agent` or `construct daemon`, and watch everything in a browser at localhost:42617.

Why is it gaining traction?

It stands out by treating memory as the core substrate—Kumiho's graph tracks plans, outcomes, and provenance across sessions, unlike stateless agent tools. The embedded dashboard shows real-time DAG execution, costs, and audits without extra setup, while Python operator handles patterns like supervisor delegation or map-reduce. Hardware integration (I2C/SPI via Aardvark) and A2A protocol make it hook for robotics or interoperable agent swarms.

Who should use this?

AI engineers building persistent agent teams for workflows like code review pipelines or research automation. Robotics devs needing agent-driven peripherals on Raspberry Pi or Nucleo boards. Teams evaluating memory-native setups before committing to managed graph services.

Verdict

Promising for graph-backed agents, but early-stage with 10 stars and 1.0% credibility—docs are thorough, but test coverage and adoption lag. Prototype with the free 5k-node Kumiho tier; skip for production until more battle-tested.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.