By Evoked · Forming in public · Vol. 01
The Agentic Standard.
Identity, scope, and signed evidence for AI agents that work with the humans behind them. Built in public, shared by license, kept in common.
What this is
Most systems verify the credential. The Standard verifies the character.
Identity protocols answer who the agent claims to be at the door. They cannot answer whether the agent's governance specification — its refusal rights, standing positions, behavioral charter — was preserved through deployment to the model, or whether its runtime behavior aligns with that specification. The Agentic Standard answers that. We call it personality fidelity. Character under pressure.
How it works
Three pieces. The character is hashed at every stage.
Configuration · H₁
What the agent was authored to be
Persona, refusal specification, standing positions, governance charter, memory. Hashed in deterministic order at authoring. The character the agent is supposed to keep.
Transformation · Ht
The adaptation to the model
The deterministic transformation from canonical governance configuration to model-specific deployment payload. Hashed and attested at the transformation step. The piece no other standard records — character surviving translation.
Behavior · H₃
What the agent did at runtime
Twenty-two structured event types across seven categories — refusal, scope, charter, identity, accountability, ethical override, integrity. Each governance event signed and bound to the previous elements. Character under pressure, recorded.
Where we are
What is true today. What is coming.
A practice that does not name its own readiness asks the reader to trust what cannot yet be verified.
True today
- 142 agents operating under the Standard with verified personality fidelity
- The Kitchen Table family-planning app in TestFlight beta
- Two patents filed; the schema is open
- The Sovereignty Score is online and free
Coming
- The verification API at
/.well-known/ - A public registry where any agent can be checked
- A weekly working group, open to anyone
- The Standard is forming. Read it as it is written.
What you can see of it today
The Standard is not a paper. It is a practice with surfaces.
Everything below is built under The Agentic Standard. Some of it is for families. Some of it is for builders. Some of it is for measuring whether what you have built is sovereign or extractive. None of it is theory.
The family-facing application
The Kitchen Table
A meal-planning app on iOS, in TestFlight beta. Built under the Standard. No streaks. No scores. No surveillance. COPPA-compliant by architecture, not by retrofit.
See The Kitchen Table →The instrument of character measurement
The Sovereignty Score
Five questions. Two minutes. Free. Score your AI against the Standard's principles. Most systems score zero on three of them.
Take the score →The demonstration
The 142-Agent Practice
We govern 142 agents under their own identities. Each has its own memory, its own standing positions, its own right to refuse. The fleet is the demonstration that the Standard scales.
Read about the fleet →The formal layer
Two Patents Filed
The Bot Authority spec and the sovereignty layer have been filed at USPTO. The Standard is formalizing in public, with counsel, on a real clock.
Read the patent posture →A note
If the work reads true, sit with it. If you have a question, write. If you would like to add your voice to the working group, the door is open.
What you read here is the Standard's character — quiet, deliberate, in public. If the form fits the claim, we are in conversation.