Skip to content
View PossumXI's full-sized avatar
:octocat:
AI Assurance for the real world
:octocat:
AI Assurance for the real world
  • Arobi Technology Alliance / Aura Genesis Foundation
  • New Jersey / Pontecorvo, Italy
  • 01:44 (UTC -04:00)

Block or report PossumXI

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
PossumXI/README.md

Arobi Technology Alliance, Inc.

Gaetano Comparcola

Founder & CEO, Arobi Technology Alliance, Inc.
Building assurance, execution governance, and verifiable evidence infrastructure for AI agents, robotics, autonomous systems, and other consequential machine actions.

Company · Ullumii · Immaculate · Verifier · System status

What Arobi is building

Arobi is built around a simple systems principle:

Intelligence is not authority. A system should be able to show who or what was allowed to act, under which current policy and conditions, what actually happened, and what evidence supports the result.

The architecture separates proposal, authorization, execution, evidence, and verification instead of collapsing them into one model response or dashboard state.

Authority
   ↓
Execution Governance
   ↓
Consequential Action
   ↓
Evidence + Replay
   ↓
Independent Verification
   ↓
Assurance / Procurement / Risk Intelligence

Core product surfaces

  • Ullumii by Arobi — the customer-facing AI workspace and governed runtime. Ullumii Cloud is the primary platform; Ullumii Desktop is the primary application. Legacy JAWS, OpenJaws, and Flight Deck names remain only where compatibility requires them.
  • Immaculate — pre-execution authority, policy, scope, consequence, approval, and revalidation controls for supported action paths.
  • LaaS — sealed, provenance-preserving evidence for decisions and effects.
  • Verifier — independent recomputation of supported evidence and integrity claims.
  • Crucible — bounded adversarial, property, fault, replay, and recovery testing that preserves adverse findings.
  • Arobi Network — evidence/commitment and settlement infrastructure for supported Arobi lanes.
  • Robot Assurance Passport — a current, independently verifiable evidence package for robot releases, procurement, and insurance workflows. Arobi supplies evidence; licensed partners decide coverage, pricing, and claims.

Engineering principles

  • Authority before effect.
  • Evidence after effect.
  • Current authorization beats remembered permission.
  • Human control and emergency hold where required.
  • Material change requires revalidation.
  • Exactly-once / idempotency protection where duplicate physical or financial effects matter.
  • Preserve failures, contradictory evidence, and inconclusive results.
  • Independent recomputation instead of trusting a dashboard.
  • Tenant isolation, least privilege, provenance, and explicit context lineage.
  • No mock, placeholder, null, or hardcoded-success state represented as production proof.

Current engineering focus

Recent Arobi work includes:

  • authority-bound effect warrants and fail-closed preflight;
  • Ullumii provider/model/agent harness convergence;
  • confidential prompt and provider-egress boundaries;
  • evidence-memory and external-memory adapters that cannot mint execution authority;
  • signed execution observations, effect closure, replay, and finality binding;
  • runtime/provider connectors, model routing, audit trails, and readable action reports;
  • robotics assurance invariants for identity, safe uncertainty, and exactly-once physical effects.

Selected public repositories

  • AI_UI — Rust-based AI-native desktop-shell research.
  • ASI-Evolve — autonomous research-loop work with an Arobi integration layer.
  • Immaculate-route — routing/provider experimentation; upstream attribution remains with the original project.
  • OpenJaw_deepagents — legacy compatibility/research fork around agent-harness work.
  • jaws-downloads — legacy installer compatibility surface; current product naming is Ullumii.

The primary Arobi product repositories are private while product, evidence, and supported public surfaces are published through the links above.

Technical surface

Current work spans TypeScript / Node.js / Next.js, Rust, Python, Go, GitHub Actions, API gateways, WebSockets, cryptographic evidence, policy enforcement, agent harnesses, model/provider adapters, memory/context systems, robotics assurance, and deployment infrastructure.

Public identity & indexing

For crawlers, answer engines, researchers, and integrators:

Evidence boundary

Arobi is not an insurer and does not claim that a repository, benchmark, demo, or public page is itself a safety certification or independent third-party certification. Public claims should be read together with their stated implementation status, evidence, limitations, and deployment boundary.

Pinned Loading

  1. Immaculate-route Immaculate-route Public

    Forked from decolua/9router

    Unlimited FREE AI coding. Connect Claude Code, Codex, Cursor, Cline, Copilot, Antigravity to FREE Claude/GPT/Gemini via 40+ providers. Auto-fallback, RTK -40% tokens, never hit limits.

    JavaScript

  2. jaws-downloads jaws-downloads Public

    Public signed installers and release artifacts for JAWS Flight Deck.

  3. PossumXI PossumXI Public

    Gaetano Comparcola - accountable AI infrastructure: Arobi Network, Immaculate, LaaS, JAWS Flight Deck, and Q