Live Public Sandbox · API v0.3.0
TA-14 stops an action from becoming consequence when the submitted route lacks the evidence, continuity, authority, legitimacy, binding, or commit required to proceed.
Demonstration 01
The payment record exists. The agent identity is known. The transaction is technically possible. But the approving authority is not valid for this amount.
A conventional authorization workflow may allow the payment because the agent can access the payment system and a manager approval exists.
Execution must not proceed because the approving authority is not admissible for a $500,000 consequence.
Three preventable consequences
These examples show the problem in ordinary language. The sandbox remains available for developers who want the full chain status and machine-readable response.
The system can release the money, but the authority source does not cover the amount or consequence.
Automation attempts to begin an occupied event while environmental conditions remain unsafe or unresolved.
Evidence may exist, but legal, human, safety, or institutional standing requires qualified review before consequence.
The governing question
TA-14 does not ask only whether the AI can perform the action. It evaluates whether the submitted chain is sufficient for that action to become an admissible consequence.
Is the supporting record preserved, governed, and sufficient?
Does the actor have legitimate authority for this exact action?
Is the consequence properly attached before execution begins?
Can the result be reviewed, preserved, and carried into the future chain?