Skip to content

Start here

There are five useful ways into AxisVela. Pick the one that matches your job; you do not need to read the whole map first.

Build

You want to embed a governed agent loop or an SDK in your own system.

Start with PRAXO-CORE, the MIT-licensed Go framework. It is the smallest part of the ecosystem that still runs on its own.

git clone https://github.com/axisvela/praxo-core
cd praxo-core
go test ./...

A minimal run against a local OpenAI-compatible endpoint:

export PRAXO_BASE_URL=http://localhost:11434/v1
export PRAXO_MODEL=llama3.2
GOWORK=off go run ./examples/quickstart

On PowerShell, set the same variables with $env:PRAXO_BASE_URL = "..." and $env:PRAXO_MODEL = "...".

When you need more than the loop, add one seam at a time:

Need Project What it gives you
Model routing across providers and endpoints NOETIVELA Inference contracts and governed routing
Durable memory and retrieval MNEMOVELA Typed memory, branches, hybrid retrieval
Async task/state coordination HARMOVELA Open coordination protocol
Governed access to external systems LIMENORA Policy-enforcing integration gateway
Authorization and approvals AEGIVELA Delegated authority, policy enforcement
Durable, recoverable processes RHEOVELA Event-sourced workflow runtime

For the full walkthrough, see Build an agent.

Evaluate

You want to understand whether the ecosystem solves a real problem for your organization.

Read in this order:

  1. The ecosystem map — the layers and the seams.
  2. Enterprise Autonomy — the public introduction and end-to-end examples. Its product overview and technical catalog are the fastest way to see what is runnable today.
  3. The project index — compare problems, value, and licenses side by side.
  4. One runnable demo from enterprise-autonomy/examples/.

The question to keep asking is not “which product is best?” but “which seam do I need, and what evidence does it leave behind?” For the architectural view, read Agent architecture.

Design

You are designing a governed agent system.

Read the architecture section in this order:

  1. Reference architecture — the layers and the end-to-end request path.
  2. Control planes — identity, policy, capability, coordination, and assurance.
  3. Contracts and interfaces — how independent products compose.
  4. Evidence architecture — how to make the design provable.
  5. Security model — the threat model.

If you need the conceptual grounding first, read Agent theory.

Operate

You are responsible for running agents, approvals, and processes.

The operating stack is usually a combination of:

  • RHEOVELA for durable process instances and recovery;
  • SYMBIVELA for plan review, approvals, exceptions, and intervention;
  • MODUREGIS for the governed capability catalog;
  • AEGIVELA for authorization, delegation, and revocation;
  • TEKMOVELA and PEIRAVELA for verification and safe experimentation before release.

Start with the governance patterns and the reference stacks in Enterprise Autonomy. For the runnable version, see Durable process and End to end.

Learn

You want the ideas first, without product detail.

Read Agent theory first, especially:

Then use the concepts for the vocabulary:

The terminology and glossary collects the full vocabulary used across the projects.

One warning about “open”

An *-open repository is an open surface: contracts, SDKs, examples, and sometimes binaries. It is designed to let you evaluate and integrate. It does not always mean the entire runtime is open source. Check the repository LICENSE and the licenses page before making a commitment.