Arkavo / developers

Developer tools from Arkavo

Agents on your hardware.
Protection on your data.

Run agent workflows locally with access policies you configure. Protect the data itself with OpenTDF; we maintain open-source implementations in Swift and Rust. Render native avatars with Metal.

Get started with EdgeRender with VRMMetalKit

Three tools. Distinct jobs.

Available · Open source

Arkavo Edge

Run and coordinate agent workflows with local models, optional cloud providers, and configurable tool access.

Explore Arkavo Edge

Open-source security & interoperability

Security that travels with the data.

Access rules usually live in the system holding the data. OpenTDF binds them to the data itself, so protection follows a file, a model, or an agent’s output wherever it goes.

Swift · Rust

OpenTDF implementations

Community-driven, open-source OpenTDF libraries for Apple platforms and Rust.

OpenTDFKit · opentdf-rs

Agent2Agent protocol

A2A SDK and interop matrix

A Swift A2A SDK, plus a pairwise conformance matrix that tests real SDKs against each other and captures wire evidence for every failure.

a2a-swift · a2a-conformance

Formal verification

TØR-G

A zero-parser, boolean-circuit representation for AI agent policies that can be formally verified.

torg on GitHub

Start with a running agent.

Arkavo Edge

Install the release for your platform, start an agent, and explore its web interface. For a concrete next step, inspect the code-review example and its role-specific access policies.

Installation & first runCode-review example

After installation

arkavo

# Or start the web interface
arkavo ui

First launch downloads local models.

Understand the boundaries.

Local inference, provider requests, and tool access are different parts of a workflow. Know what runs where before choosing a configuration.

Read the trust & data guide

From the engineering journal

Can a model narrow access without granting it?

An experimental OpenTDF case study: the deny-only boundary, a threshold that suppressed the intended restriction, and what still needs evaluation.

Read the case studyExplore the journal

More open-source work.

arkavo-edge

Agent workflows with local models, optional cloud providers, and configurable access policies.

game-rl

Multi-agent AI infrastructure for games. Turn any game into a multi-agent environment — train RL policies, orchestrate LLM-powered NPCs, or build AI game masters.

WeDLM-rs

Rust implementation of Tencent's WeDLM parallel decoding architecture for fast LLM inference.

Explore arkavo-ai on GitHub · Explore arkavo-org on GitHub