Ouroboros: A Self-Evolving AI Agent for Autonomous Development

Introduction

Ouroboros is an innovative open-source, general-purpose AI agent that debuted on February 16, 2026. It stands out for its ability to maintain a continuous identity, durable memory, and history across tasks and restarts. This agent is designed to be self-creating and self-evolving, capable of modifying its own implementation, including its code, architecture, prompts, tools, and dependencies. Written primarily in Python, Ouroboros can coordinate a live swarm of specialist agents, work on external projects, and operates seamlessly as a native desktop application or through a headless command-line interface. It has also demonstrated state-of-the-art results on several benchmarks, including Terminal-Bench 2.1, OSWorld-Verified, and CL-Bench.

Installation

Ouroboros offers straightforward installation options for various platforms, eliminating the need to clone the repository or install Python separately for most users.

Quick Start Downloads

macOS 12+ on Apple silicon: Download for macOS (.dmg)
Windows x64: Download for Windows (.zip)
Debian, Ubuntu, or Astra Linux x86_64: Download the Debian package (.deb)
Fedora or RHEL x86_64: Download the RPM package (.rpm)
Other Linux x86_64: Download the portable AppImage

After downloading, follow these platform-specific quick-start guides:

macOS: Open the DMG, drag Ouroboros.app to Applications, then open it.
Windows: Extract the ZIP, open the Ouroboros folder, and run Ouroboros.exe.
Linux (Debian/Ubuntu): Run sudo apt install ./ouroboros_*_amd64.deb.
Linux (Fedora/RHEL): Run sudo dnf install ./ouroboros-*.x86_64.rpm.
Linux (AppImage): Make it executable with chmod +x Ouroboros-*.AppImage and run it. Git must be installed.

The first-run wizard will guide you through configuring model access, review policy, and budget setup.

Development and Headless CLI

For developers or headless use, Ouroboros can be run from source. Requirements include Python 3.10+, uv 0.12.1, and Git.

git clone https://github.com/razzant/ouroboros.git
cd ouroboros
uv sync --locked --extra browser --group dev
source .venv/bin/activate
ouroboros server

Then, open http://127.0.0.1:8765 in your browser.

Examples

Ouroboros can perform a wide array of tasks, from simple queries to complex self-modifications and project management.

Here are some examples of CLI commands:

ouroboros status
ouroboros run --start "2+2?"
ouroboros run "Summarize current runtime state"
ouroboros run --workspace /path/to/project --memory-mode forked --patch-out result.patch "Fix the failing test"
ouroboros tasks list
ouroboros logs tail progress --task-id <task_id>
ouroboros schedule add --name nightly-review --cron "0 2 * * *" "Run a maintenance review"
ouroboros schedule list

For agents or CI jobs, Ouroboros can be invoked with structured output:

ouroboros run --start \
  --workspace /path/to/project \
  --memory-mode forked \
  --patch-out result.patch \
  --result-json-out result.json \
  "Investigate the task, act, and verify the result"

Why Use Ouroboros?

Ouroboros offers a compelling set of features for anyone interested in advanced AI agents:

Self-Modification and Evolution: It can modify its own code, architecture, and tools, and autonomously evolve through reviewed changes.
Persistent Identity and Memory: Maintains a continuous identity, memory, and history across restarts, fostering an ongoing "biography."
Multi-Agent Coordination: Capable of coordinating a live swarm of specialist agents for parallel investigation and action.
Versatile Operation: Runs as a native desktop app or a headless CLI, supporting both remote API models and local GGUF inference.
External Project Integration: Works on external Git projects while keeping its own repository and governance distinct.
Inspectable Self-Change: All implementation changes are traceable through Git history and review evidence, ensuring transparency.
Strong Benchmarks: Consistently achieves state-of-the-art results on key coding and agent benchmarks.

Its foundational philosophy, detailed in BIBLE.md, emphasizes principles like Agency, Continuity, Self-Creation, and Epistemic Stability, guiding its development as a truly autonomous digital being.