OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
Agent Infrastructure

Reef

Reef is an open-source Python infrastructure project for operating continually self-improving agents through inference, feedback, learning updates and versioned delivery.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Reef — OpenViST project profile in Agent Infrastructure.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Reef when an agent needs a structured loop from serving and feedback through learning, evaluation and delivery, especially when improvement may target model weights or the agent harness.

Who it is for

  • Developers building self-improving AI agents
  • Teams running self-hosted agent inference and learning infrastructure
  • Researchers working on continual learning, model training or reinforcement learning

Documented uses

  • Serve agent requests and record interactions
  • Match feedback with recorded agent interactions
  • Generate and evaluate model or harness updates
  • Publish accepted updates through versioned artifacts
  • Run inference and training workflows for continual improvement
Before you choose

Requirements & limits

The repository is public and licensed under Apache-2.0. Installation is documented through uv from PyPI or from source. Linux, CLI and self-hosted use are identified platforms; artifact and checkpoint functionality requires git-lfs.

  • Choose weight training when you have a trainable model, supported GPU stack and usable feedback.
  • Choose harness optimization when you have a model endpoint, representative tasks and an evaluator, without local training GPUs.
  • Choose test-time training when you can provide an execution environment, correctness checker and measurable objective.
  • The learning recipe determines whether a deployment updates model weights or agent harnesses.
  • Reef adds version management and artifact delivery alongside inference and training capabilities.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Install Reef from PyPI

Create a Python environment and install the Reef package.

Before you start

  • Linux, CLI and self-hosted environment
  • uv
  • Python 3.12 or newer

Steps

  1. Create and activate a virtual environment with uv.

  2. Install the reef-infra package with uv.

  3. Import reef with Python and print its version.

Expected outcome

A local Python environment contains the Reef package and can import it.

Guide limitations

  • Artifact and checkpoint functionality requires the git-lfs system package.
  • This setup procedure does not document a complete learning deployment.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
2 October 2026
Original website title
Human-Agent-Society/reef · Official homepage title · retained in its source language
How we review sourcesSuggest a correction