Assertion-Cost Symmetry

Making an accusation costs the accuser too — symmetric, metered, replay-proof.

Primary technical disclosure

Secondary technical

Withheld Delegating Faculty, Preserved Direct Execution How a governed agent withholds only its faculty to delegate an action class while preserving direct execution, when a residue-surcharged accepted determination drives its deviation quantity across the policy bound.Depth-Propagated Harm Inheritance Across a Delegation Chain How the Adaptive Query architecture propagates a returned harm-recording field class up a multi-hop delegation chain so harm caused by a sub-delegate at any depth is inherited and priced at every delegating agent above it.Co-Signature Conditioning of Restoration Authority How a governed autonomous agent meters its own principal's restoration authority and conditions a repeat return of the authorization gate on a co-signature from a governance address named in the principal's own policy.The Anti-Evasion Charge on Governance-Address Removal How the Adaptive Query governance architecture prices, rather than honors, an autonomous agent's attempt to drop a named governance address in order to evade a subsisting co-signature requirement on restoration authority.The Progressive Issuance Decrement How progressive issuance decrement prices the breadth of an AI agent's accusations by making the authorization-budget decrement rise with the count of distinct recipient origin-equivalence classes addressed within a metering window.Non-Attenuable Forward Restraint Inheritance: A Delegation Record That Cannot Strip an Agent's Abstention State How a delegation record's forward capability mask is tested on append against a policy-declared non-attenuable forward field class, so an agent cannot hand a delegate a cognitive state stripped of the scoped integrity vector, refusal counter, and authorization gate whose presence conditions abstention.Holding Authorization When the Renewal Register Cannot Be Read: Re-Read Cadence, Held-Value Resume, and the Max-Hold Inquiry How a persistent semantic agent that cannot read its authorization-renewal register holds the authorization quantity at its current value, re-reads the register on a policy-declared cadence, resumes decay from the held value, and emits a structured inquiry to its principal once a declared maximum hold interval has elapsed.

Applications · general

Applications · specific

Terminology