AxisVela Open

A public map of independent projects for governed autonomy.

English | 中文

Documentation version: v0.2.0

AxisVela is an independent, open-source project family for people building and operating agents, workflows, robots, digital twins, and the systems that connect them. The projects share a design stance — contracts over a shared kernel, evidence over assertions, and human control over blind automation — but each product is independently versioned and can be used on its own.

This repository is the map, not a monorepo. It explains what each project is for, the problem it solves, the value it provides, and how the pieces relate.

Start here · Products · Theory · Architecture · Practice · Concepts · Reference · Machine-readable index

Start with the problem, not the product

If you need to…	Start with
Build an agent loop you can govern and embed	PRAXO-CORE — MIT Go framework
Understand how the whole ecosystem fits together	Enterprise Autonomy — public introduction and runnable examples
Run a long-lived, recoverable business process	RHEOVELA — durable workflow platform
Give agents verifiable identity and bounded authority	EIDOVELA, NOMIVELA, AEGIVELA
Connect agents to APIs, MCP servers, and events without bypassing policy	LIMENORA — governed integration gateway
Design and validate enterprise architecture with AI assistants	TAXOVELA — architecture skill pack and CLI

What “governed autonomy” means here

Governed autonomy is not a model feature. It is an operating discipline built from three questions:

1. What is allowed? Capabilities, data, and side effects are bounded by explicit policy before execution.
2. What actually happened? Actions leave evidence: effect ledgers, traces, process history, simulation records, or verification results.
3. Who decided? High-risk steps have a named owner, an approval path, and a way to intervene, revoke, or recover.

Different projects answer these questions at different layers. The concepts section explains the ideas; the ecosystem map shows where each project sits.

The map at a glance

Layer	What it covers	Representative projects
Design and standards	Architecture design, validation, and portfolio management	TAXOVELA, PAMP
Foundation	Agent framework, coordination protocol, development coordination	PRAXO-CORE, Harmovela, COORDIVELA
Runtime and cognition	Agent execution, memory, inference routing	PRAXOVELA, Mnemovela, Noetivela
Work and capabilities	Planning, capability governance, durable process	ORCHADYN, MODUREGIS, RHEOVELA
Trust and knowledge	Identity, registry, authorization, grounding, world models	EIDOVELA, NOMIVELA, AEGIVELA, GNOSIVELA, ONTOVELA
Boundary and assurance	Integration, verification, simulation, skill supply chain	LIMENORA, TEKMOVELA, PEIRAVELA, SYNDOVELA
Human and physical operations	Human oversight, robot fleet control	SYMBIVELA, KINETOVELA
Program	End-to-end composition and examples	Enterprise Autonomy

The layers are a reading aid, not a dependency chain. Products integrate through versioned contracts and can be adopted independently.

Open surfaces and commercial editions

• Fully open projects: PRAXO-CORE (MIT), TAXOVELA (MIT), and the Harmovela protocol (Apache-2.0).
• Open product surfaces: most products publish an Apache-2.0 *-open repository containing contracts, SDKs, examples, and sometimes binaries. The runtime or enterprise edition may be developed separately.
• Licenses vary: always treat the LICENSE in the specific repository as authoritative. See licenses for the current map.

The practical consequence: you can evaluate the interfaces and build against them without buying anything, but an open surface is not a promise that the entire runtime is open source.

Suggested reading order

If you are…	Read
New to the ecosystem	Start here → Ecosystem map → Products
An architect	Agent theory → Agent architecture → Enterprise Autonomy
A builder	Build an agent → Govern an agent → Durable process
A researcher	Agent theory → Research map → Reading list

How this repository is organized

• docs/projects/ — one page per project: problem, what it provides, why it matters, and how it fits.
• docs/theory/ — the conceptual spine: UAD, DIKCA, intent/AOP, PPES, taxonomy, multi-agent dynamics, economics. It synthesizes Huiwen Han’s research program and links each page to the original source.
• docs/architecture/ — runtime, control planes, contracts, knowledge, evidence, deployment, and security.
• docs/practice/ — step-by-step guides using AxisVela as the reference implementation.
• docs/concepts/ — the vocabulary and the deeper conceptual notes, including a comprehensive terminology section.
• docs/reference/ — relationships, licenses, sources, standards, and FAQ.
• projects.yaml — a machine-readable project index for tooling, catalogs, and AI agents.
• llms.txt — a compact index for AI-assisted discovery.

Status

These projects are early stage. Public contracts are versioned, but APIs may change between minor releases until a repository declares 1.0. Each project’s CHANGELOG.md, ROADMAP.md, and release tags are the source of truth.

AxisVela is a solo-maintainer, independent project family — not a company. The public repositories are the open surface; commercial and enterprise editions are developed separately.

Contributing, security, and license

• Contributing — documentation fixes, project proposals, and the boundary between this repo and product repos.
• Security — report vulnerabilities privately.
• Code of Conduct.
• Documentation and examples in this repository are licensed under Apache-2.0. Product repositories carry their own licenses.