Skip to main content

​
Build Goal-Driven AI Agents

Aden Hive is a framework for building production-grade AI agents without manually wiring rigid workflows. You define outcomes in natural language, and Hive helps generate and run node graphs that can adapt when failures happen.

Build Your First Agent

Create a goal-driven agent in minutes with Claude Code skills.

Core Concepts

Understand goal-driven development and agent architecture.

​
Why Aden?

Traditional agent frameworks require manual workflow design, brittle control flow, and reactive fixes. Aden flips this model: you describe outcomes, and the framework helps build and evolve the system.

Goal-Driven

Define objectives in natural language; the coding agent generates graph structure and connection logic.

Self-Improving

Runtime failures become inputs for iterative improvement and graph evolution.

Human-in-the-Loop

Built-in intervention nodes pause execution for human input with configurable timeouts.

Production-Ready

SDK-wrapped nodes include shared memory, observability, tool access, and control hooks.

​
Quick Start

# Clone and run setup wizard
git clone https://github.com/adenhq/hive.git
cd hive
./quickstart.sh

# Build an agent using Claude Code
claude> /hive

# Test your agent
claude> /hive-test

# Run your agent
PYTHONPATH=exports uv run python -m your_agent_name run --input '{...}'

​
Run an Existing Agent

You can also run an existing agent from exports/ (or copy one from examples/ first). 
PYTHONPATH=exports uv run python -m existing_agent_name run --input '{...}'
For full reuse/refactor workflow, see Build From Existing Agent.

Quickstart Guide

Complete setup instructions and your first agent in under 10 minutes.

​
What Makes Aden Different

Traditional FrameworksAden
Hardcode agent workflowsDescribe goals in natural language
Manual graph definitionAuto-generated agent graphs
Reactive error handlingOutcome-based adaptation
Static tool configurationsDynamic SDK-wrapped nodes
No native intervention modelBuilt-in HITL and control points

​
Platform Components

Agent Building SDK

Build agents with nodes, edges, and goal-driven development.

Core Concepts

Learn goals, graph runtime, worker sessions, and evolution loops.

Credentials

Configure secure credential storage, refresh, and sync patterns.

Testing Framework

Validate agent behavior with goal-based testing.

​
Need Help?

Book a Discovery Call

Schedule a call with our team to learn more.