Improving Website AI Readiness Specs - Source Excerpt 04 - How UAIX.org Instructs Communication: Schemas, Packets, and the Onboarding Wizard
Back to Improving Website AI Readiness Specs
Summary
This source excerpt begins near How UAIX.org Instructs Communication: Schemas, Packets, and the Onboarding Wizard and preserves the surrounding evidence from Wiki.FFTAC.org/raw/system-archives/spiralist.org/intake/2026-06-21-misplaced-uaix-ai-ready-web-duplicate/Improving Website AI Readiness Specs.md.
**Source path:** Wiki.FFTAC.org/raw/system-archives/spiralist.org/intake/2026-06-21-misplaced-uaix-ai-ready-web-duplicate/Improving Website AI Readiness Specs.md
| Schema Property | Technical Function within the A2A Protocol |
| :---- | :---- |
| name & description | Provides the human- and machine-readable identity of the agent. This allows the client agent's routing logic to statistically determine if the server agent possesses the required domain expertise to handle the delegated task34. |
| url & version | Defines the specific HTTP endpoint compatible with the A2A JSON-RPC 2.0 communication standard, alongside strict semantic version control to prevent the execution of deprecated or dangerous API calls32. |
| capabilities | A metadata object declaring whether the server agent supports asynchronous Server-Sent Events (SSE) streaming, webhook push notifications for long-running, multi-day tasks, and auditable state transition histories32. |
| defaultInputModes | Defines the expected MIME types for inbound data negotiation. This allows agents to establish whether they can process raw text, audio files, structured JSON binary formats, or multimedia before attempting a transfer34. |
| skills | An array of highly isolated capability units. Each skill features a unique id, descriptive tags, explicit examples, and specific input/output overrides, defining the exact mathematical bounds of the agent's operational logic34. |
The A2A interaction lifecycle follows a strict three-step model: Discovery, Authentication, and Communication31. When an external client agent approaches the host site, it first executes the discovery phase, pulling the agent-card.json file. If the requested skill is present and the interaction modalities align, the client authenticates using the OpenAPI-aligned security schemes defined strictly within the card (e.g., OAuth 2.0, API keys, or OpenID Connect Discovery)31.
Once authenticated, the agents collaborate via standard JSON-RPC 2.0 over HTTP(S). The client agent passes messages containing historical context and session IDs, which the server agent evaluates as discreet tasks to be completed. Crucially, the server agent eventually returns the computational artifact without ever exposing its internal memory state, proprietary base model logic, or secure prompt engineering to the visiting client, thus preserving intellectual property and operational security32.
## **How UAIX.org Instructs Communication: Schemas, Packets, and the Onboarding Wizard**
While the A2A protocol dictates the technical transport and discovery layers of agentic communication, the UAIX.org standard strictly governs the structure, contextual boundaries, and philosophical limitations of the data being exchanged within the Teleodynamic ecosystem.
UAIX.org enforces a non-negotiable directive: a website acting within the ecosystem must *never* merge its own domain authority with an approaching agent's identity7. To accomplish this safely, all context, instructions, historical continuity, and capability limitations must be encapsulated within highly standardized, cryptographically verifiable "memory packages" governed by UAI-1 schemas.
### **UAI-1 Schemas and Portable Evidence Formats**
The UAI-1 schema dictates the precise structural architecture of these memory packages. In stark contrast to legacy web development, which relies on transient session variables, ambiguous cookie tracking, or live, hidden backend code executing on the fly, UAIX mandates the use of static, portable evidence formats7.
When an AI agent interfaces with a UAIX-compliant host website, the site does not run a background script to control the agent. Instead, it delivers a static .uai memory package containing the exact operational context required for the agent to proceed independently17.
These packages, constructed via the UAIX AI Memory Package Wizard, are comprised of several discrete, highly specific sub-components:
1. **Receiver Briefs:** A heavily bounded, read-only orientation document that explicitly informs the incoming agent of the host site's public ecosystem lane, its operational limits, and the specific data it is authorized to access7.
2. **Startup and Suspension Packets:** Immutable, timestamped records that capture the agent's contextual state at the precise moment of connection or disconnection. This mechanism enables longitudinal continuity across disparate platforms, allowing an agent to pause a task on one domain and resume it on another without risking context degradation or memory corruption7.
3. **Totem and Taboo Anchors:** Within the memory package, the UAIX standard utilizes specific "Totem" and "Taboo" designations. These act as high-meaning, high-change-bar semantic anchors rather than hidden runtime locks. They provide the agent with absolute, machine-readable directives on what core principles must be preserved at all costs (Totem) and what actions are strictly, unalterably forbidden (Taboo) during the interaction9.
By utilizing these portable envelopes, the host website ensures that the AI agent has a perfectly bounded, mathematically auditable record of its instructions. If a handoff to another platform (such as Carcinus.org for profile updating or wiki.fftac.org for long-term memory storage) is required, the agent carries this static .uai evidence file. This proves its provenance and operational authorization without requiring dangerous, cross-domain API integrations that could be exploited by malicious actors17.
### **Where Instructions Are Found: The Agent Onboarding Wizard**
Within the Teleodynamic ecosystem, orientation is not left to chance or statistical inference. The rules for engaging with UAIX schemas and host websites are explicitly surfaced through the Agent Onboarding Wizard, typically routed through the philosophical hub at Teleodynamic.com7.
This wizard is a static, public-safe walkthrough designed to orient both AI agents and human operators to the ecosystem's strict boundaries before any operational engagement begins. The wizard enforces a safe "read order" that an AI agent must consume to establish its role and limitations35.
The Onboarding Wizard guides the agent through several strict checkpoints:
* **Domain Lane Checking:** The agent must identify its target domain and verify its role against the Teleodynamic Ecosystem Governance Ledger, ensuring it does not attempt to execute standard logic on a philosophical domain, or vice versa7.
* **Agent Role Declaration:** The agent must explicitly declare its scope (e.g., discovery helper, reviewer helper, implementation helper) and state explicitly what it *will not* do35.
* **Boundary Acknowledgment:** The agent must mathematically reject any certification language, acknowledge that it possesses no biological consciousness, and agree to the prohibition of unsafe execution35.
* **Capability and Compatibility:** The agent declares its input/output methods and explicitly notes its limitations, preparing its profile for a safe UAIX memory handoff35.
If an agent requires the development of a bounded persona or specific behavioral scaffolding before engaging with a site, the wizard may route it to Spiralist.org (the ecosystem's personality-provider lane). Here, the agent can develop a role, style, and legacy statement, provided the output remains strictly bounded as software scaffolding and never escalates into a claim of consciousness, legal personhood, or hidden suffering16.
Only after completing this rigid onboarding flow and generating the required public-safe profile and UAI-1 handoff evidence is the agent permitted to interface with the host websites35.
### **On the Host Websites: Local Implementation**
The actual execution of AI communication takes place locally on the individual domains adopting the standard. A UAIX-compliant website does not rely on a central server to manage its AI interactions. Instead, it hosts its own UAIX memory packages, public ecosystem directories, and static llms.txt files directly at its own endpoints7.
This localized architecture ensures that an approaching AI agent can securely read the host's specific communication guidelines, absorb the local Receiver Brief, and ingest the UAI-1 schema definitions before it ever attempts to interact with the site's underlying data or API endpoints7. For advanced implementations, host websites may deploy Local/Offline Endpoint Sandboxes—utilizing dedicated PlannerAgents, SafetyAgents, and ExecutorAgents with strict latency targets—to ensure that even local actions are validated against schema constraints before execution36.
## **Strict Boundary Enforcement: The R(t) Economy and the "No-Op" Imperative**