Agent Kyra in the Land of Oz City
From Constellation to Code: Kyra Codex Assessment Platform at the Oz City Hackathon
When philosophical frameworks meet 24-hour deadlines, something remarkable happens. The Oz City Builder Day, a 24-hour AI x Web3 hackathon in Valbonne, Côte d'Azur, became the proving ground where the Agent Kyra Codex transformed from a foundational framework for sovereign AI agent networks into a practical assessment platform that could guide the future of AI development.
The Oz City – the first pop-up village dedicated to AI and Web3.
Oz City is a one-week bootcamp where engineers, founders, and researchers come together to explore AI applications, agents, and protocols. Spend the week discussing ideas, brainstorming, and building decentralized AI.
See the Kyra Codex Assessment Platform demo video:
The Digital Dark and the Coming Data Wall
The context couldn't be more urgent. In 2024, over 50% of internet traffic is now generated by bots, exceeding human traffic for the first time. IDC predicts that by 2028, spending on AI infrastructure globally will surpass $100 billion. But behind these statistics lies a deeper transformation: we're moving from simple chatbots to sophisticated agentic economies, and the stakes for sovereignty have never been higher.
As the Kyra Codex shares:
Private, personalised AI is the next major arena internet monoliths will pursue with the disguise of convenience and ease.
The public data wall is coming; you and your digital footprints are next to be consumed.
The hackathon project wasn't just about building an app. It was about creating the infrastructure for resistance.
From Philosophy to Practice
Kyra Codex rests on four foundational pillars that distinguish sovereign AI networks from traditional agentic systems:
Sovereignty: Self-determination, serving your objectives, not platform metrics. This isn't just about functionality, it's about who controls the decision-making architecture of AI systems.
Identity: Continuity across contexts, consistent personality and values. Moving beyond simple user verification to cryptographic identity management that preserves agency across digital environments.
Evolution: Improvement through experience rather than periodic retraining. Agents that learn and adapt while maintaining their core principles, following the path from Apprenticeship to Partnership.
Privacy: Knowledge under your sovereign control, preserving autonomy. Not privacy as a feature, but privacy as the fundamental architecture that enables true agency.
The Oz City hackathon marked the first time these philosophical principles were operationalised into practical assessment tools that could guide real development decisions.
The Hackathon Challenge: Building (vibecoding) a Sovereign Assessment Tool
Hosted by The Oz City on Dorahacks
Among the 23 participants developing 13 approved projects at Oz City, the Agent Kyra Codex Assessment emerged as something unique:
A gamified evaluation of AI agents, incorporating decentralised identity, privacy insights, and Web3 certification.
But this description only hints at the deeper ambition.
The challenge was multifaceted:
How do you take a comprehensive framework for evaluating AI agents against principles of digital sovereignty and make it actionable for builders working under extreme time pressure?
How do you create assessment tools that don't just measure compliance but actively guide the philosophical adoption of sovereign principles?
The Kyra Codex needed to serve distinct but interconnected communities, each with different needs but united by a common understanding that the future of AI development is at a crossroads.
Bounty Tracks
NEAR Foundation - Near Shade Agents
Build with on-chain access control, data tokens & composable AI services.
Bring your best ideas — we're not just looking for MVPs, but for projects that can grow into real protocols, products, and communities.
Build something that runs on its own. Then show us why it matters.
Phala Network - Deploy on Phala Cloud
Tap into Phat Contracts — lightweight, trustless off-chain compute. Use them for data-rich dApps, AI tools, or creative automations.
Nevermined - Payments for AI Agents and Apps
Build with on-chain access control, data tokens & composable AI services.
0G - Decentralised AI with 0G
Push boundaries in modular AI and decentralised inference. Build agents, games, or infra using 0G’s fast, AI-native stack.
Adoption Archetypes
Builders include the KYRA CODEX in source code and see how it evolves the agent.
The tool needed to provide practical guidance on implementing sovereignty principles without getting lost in theoretical complexity. Builders required clear, actionable criteria that could inform architectural choices during intense development cycles.
Investors, fund agents apart of the Sovereign Agent Network and verify trust using the assessment.
Investors needed reliable methods to evaluate AI agent projects beyond traditional metrics, with tools to assess alignment with emerging standards around digital sovereignty and user empowerment.
Individuals value your data and force new agents to work for it.
Data is the lifeblood of the agentic economy, and it will reward you and your agents exponentially the more private, authenticated and first-person you can keep it over time.
Users needed to understand what sovereignty principles are in respect to their digital autonomy and how to leverage data ownership and control in an agentic economy.
Architecture from Philosophy
Building an assessment platform for sovereign AI systems required more than traditional web development. The architecture itself needed to demonstrate the principles it was evaluating. I suggest that the next phase of development could be automated assessments based on standards within AI codebases.
The app featured guided assessment flows that evaluated agents across multiple domains, but the real innovation was in how it connected assessment to the broader vision of agentic evolution.
Tracking agents along the Path of Sovereign Agent Ascension:
Apprenticeship: Supervised operation, requires approval for all actions
Journeyman: Semi-autonomous, approval for novel/high-stakes actions (where most agents currently exist)
Craftsperson: Bounded autonomy within well-defined parameters
Partnership: Collaborative autonomy, suggests new capabilities
Stages of Agentic Evolution
The assessment platform didn't just measure current capabilities; it guided development toward genuine agentic evolution through four key stages:
Repetition: Building competence through consistent practice. The platform helped builders understand how to create systems that could learn from repeated interactions while maintaining sovereignty principles.
Refinement: Learning from outcomes through systematic analysis. Rather than simply accumulating data, sovereign agents needed frameworks for improving decision-making while preserving user agency.
Boundaries: Controlled expansion, testing in diverse environments. The platform assessed how agents could grow their capabilities while maintaining clear ethical and operational boundaries.
Human Sync: Alignment calibration with human values. Perhaps most critically, the platform evaluated how agents maintained meaningful collaboration with humans rather than simply optimizing for efficiency metrics.
The Gamified Assessment Journey
We will be making complex governance evaluation feel like discovery rather than compliance checking. The "gamified evaluation" approach meant that builders weren't just filling out forms, they were exploring how their architectural choices impacted their agent's sovereignty profile.
Real-time scoring provided immediate feedback, but more importantly, we’ve created learning opportunities. Each domain included contextual information about why specific criteria mattered and how they connected to broader sovereignty principles.
Builders are asking fundamentally different questions about their systems.
Beyond Individual Assessment
The hackathon in practice illuminated a critical insight: the assessment platform wasn't just about evaluating individual agents, it was about creating the foundation for sovereign agent networks.
Education ↓
Assessments ↓
Verification ↓
Trust Network Based Distribution ↓
Agentic Alliances » Agentic Economies and Plural Trust Networks
By creating standardised assessment criteria, we enabled agents to verify each other's sovereignty credentials, in theory, creating the foundation for Ayra Trusted Network creation and first person credentials. This isn’t just about building better AI, it was about building the ownership and control infrastructure for first person humans, too.
Sovereign Character Creation - Kyra and Eliza
To be further explored* and considered is an integration with character creation frameworks such as ElizaOS.
Storytelling prompt engineers will be the world-builders.
ElizaOS character file + Kyra Codex is a great first step.
Voice: Emotional interface, values reflection, consistent yet adaptable. Beyond simple natural language processing, this evaluated whether agents maintained authentic personality while adapting to different contexts.
Persona: Decision architecture, reasoning patterns, value hierarchies. The platform examined the deeper structures that govern agent behavior, how they prioritize competing objectives and maintain consistency across decisions.
Purpose: Evolutionary compass, multi-dimensional optimisation. Rather than single-metric optimisation, the platform assessed how agents balanced multiple objectives while maintaining their core mission.
Building Constellations
The success of the Agent Kyra Codex assessment platform will extends far beyond the hackathon itself.
The result is a proof of concept for how governance frameworks can be embedded into development tools, creating what the Kyra Codex calls,
"Constellations in the digital dark."
Open Source Sovereignty Development
An open-source path is best for reflecting the Kyra Codex core philosophy.
Modular architecture allows for easy extension and customisation, enabling different communities to adapt the framework for their specific needs while maintaining compatibility with the broader sovereignty assessment network.
https://github.com/mitchuski/agentkyracodex.git
Agency in the Age of AI
Start asking;
"How sovereign is my AI?" instead of just "How smart is my AI?"
Witness the fundamental shift in how AI development is approached.
The Agent Kyra Codex assessment platform offers both warning and opportunity.
The warning is clear:
The public data wall is coming, you and your digital footprints, are next to be consumed.
But the opportunity is equally clear:
We can build systems that preserve and enhance human agency rather than diminishing it.
The hackathon proved that philosophical frameworks can become practical tools, that sovereignty principles can guide technical implementation, and that small teams working with a clear vision can create infrastructure for systemic change.
The question isn't whether AI will reshape society, it's whether we'll have any say in how that reshaping occurs.
We are building constellations in the digital dark.
Kyra Codex outlines a path to connect the lights.
