◎ OS PUB Apache 2.0 ← All specifications

P67 — AIEP — Plausibility Constraint Declaration Format

Publication Date: 2026-02-26 Status: Open Source Prior Art Disclosure Licence: Apache License 2.0 Author/Organisation: Phatfella Ltd Schema: AIEP_OS_SPEC_TEMPLATE v1.0.1 — https://aiep.dev/schemas/aiep-os-spec-template/v1.0.1

Field of the Invention

[0001] The disclosure relates to structured metadata for plausibility constraint evaluation in machine-ingestible web artefacts.

[0002] More particularly, the disclosure concerns a plausibility constraint declaration format defining a canonical structured format for declaring plausibility constraint outcomes within AIEP machine mirror pages, enabling AI agents to evaluate structural coherence of web evidence without access to internal plausibility matrix logic.

Framework Context

[0003] This invention operates within an Architected Instruction and Evidence Protocol (AIEP) environment as defined in United Kingdom patent application number GB2519711.2, filed 20 November 2025, the entire contents of which are incorporated herein by reference.

[0004] The present invention extends deterministic canonicalisation, governance, and execution integrity mechanisms defined in the AIEP environment while remaining independently implementable as described herein.

Background

[0005] Plausibility constraints evaluate structural coherence and constitutional compatibility of evidence artefacts within the AIEP framework. The internal plausibility matrix (P03) is a confidential trade secret component. However, the outcomes of plausibility constraint evaluation must be declarable in a standardised format for consumption by third-party agents.

[0006] Without a standardised declaration format, agents cannot deterministically parse or act on plausibility outcomes, and publishers cannot produce interoperable declarations.

Summary of the Disclosure

[0011] A plausibility constraint declaration format is disclosed defining four required fields per constraint entry in the plausibility object of AIEP machine mirror pages.

[0012] Each constraint entry comprises:

(a) constraint_id: a unique identifier for the constraint within the applicable schema version; (b) constraint_class: a declared class identifier such as structural, constitutional, temporal, or domain_specific; (c) result: one of three enumerated outcome values — PASS, FAIL, or WARN; and (d) severity_level: an integer severity indicator in the range [1, 5] where 1 is informational and 5 is critical.

[0013] The plausibility field in the mirror page is an array of constraint declaration objects, one per evaluated constraint.

[0014] A PlausibilityDeclarationHash is computed as:

PlausibilityDeclarationHash = H(CanonicalConstraintArray || schema_id || schema_version)

[0015] Agents encountering a constraint with result: FAIL and severity_level of 4 or 5 must treat the mirror page as non-admissible for evidence purposes unless overridden by an explicit agent-side policy declaration.

[0016] The format declares constraint outcomes only. Internal plausibility matrix logic, constraint weights, and evaluation algorithms are not disclosed.

[0017] The technical effect is provision of a standardised, implementation-agnostic plausibility constraint outcome interface enabling structured agent consumption of structural coherence declarations without disclosure of internal matrix logic.

Brief Description of the Drawings

[0018] Figure 1 illustrates the constraint entry field structure and permitted values.

[0019] Figure 2 illustrates an example plausibility array with multiple constraint entries.

[0020] Figure 3 illustrates the PlausibilityDeclarationHash computation.

[0021] Figure 4 illustrates agent fail-closed behaviour on FAIL constraint at severity 4–5.

ASCII Drawings

Figure 1 — Constraint Entry Field Structure

   +------------------------------------------+
   |       Plausibility Constraint Entry       |
   |------------------------------------------|
   | constraint_id    : unique string ID      |
   |                   within schema version  |
   |------------------------------------------|
   | constraint_class : "structural"           |
   |                   "constitutional"        |
   |                   "temporal"             |
   |                   "domain_specific"      |
   |------------------------------------------|
   | result           : "PASS"                |
   |                   "FAIL"                 |
   |                   "WARN"                 |
   |------------------------------------------|
   | severity_level   : 1 (info) ... 5 (crit) |
   +------------------------------------------+

Figure 2 — Example Plausibility Array

"plausibility": [
  {
    "constraint_id": "struct.001",
    "constraint_class": "structural",
    "result": "PASS",
    "severity_level": 2
  },
  {
    "constraint_id": "const.003",
    "constraint_class": "constitutional",
    "result": "WARN",
    "severity_level": 3
  }
]

Figure 3 — PlausibilityDeclarationHash

   CanonicalConstraintArray  ----+
   schema_id                 ----+--> H(CanonicalConstraintArray ||
   schema_version            ----+      schema_id || schema_version)
                                               |
                                               v
                              +-----------------------------+
                              | PlausibilityDeclarationHash |
                              +-----------------------------+

Figure 4 — Agent Fail-Closed Behaviour on Critical FAIL

   Agent evaluates plausibility array
          |
          v
   Any constraint with
   result=FAIL and severity_level >= 4?
          |
     +----+----+
    Yes        No
     |          |
     v          v
   NON-      Proceed
   ADMISSIBLE with
   Fail-     evaluation
   closed
   (unless agent-side
    policy override
    declared)

Detailed Description

1. Constraint Entry Fields

[0022] Each constraint entry in the plausibility array comprises:

(a) constraint_id: unique identifier within the applicable schema version; (b) constraint_class: one of structural, constitutional, temporal, or domain_specific; (c) result: PASS, FAIL, or WARN; and (d) severity_level: integer in [1, 5].

2. Array Structure

[0023] The plausibility field in the mirror page is an ordered array. Each evaluated constraint produces one entry. The array is canonicalised for hashing using the rules in P63.

3. Agent Handling

[0024] Agents encountering result: FAIL with severity_level 4 or 5 must treat the mirror page as non-admissible unless an explicit agent-side policy override is declared.

[0025] result: WARN entries are informational and do not require fail-closed behaviour by default.

4. No Internal Logic Disclosure

[0026] The format declares constraint outcomes only. Internal plausibility matrix logic, constraint weights, and evaluation algorithms are not declared in or derivable from the format.

5. PlausibilityDeclarationHash

[0027] A PlausibilityDeclarationHash is computed as H(CanonicalConstraintArray || schema_id || schema_version).

Claims

  1. A plausibility constraint declaration format for AIEP machine mirror pages comprising: (a) a plausibility array of constraint entries each comprising constraint_id, constraint_class, result (PASS/FAIL/WARN), and severity_level integer in [1, 5]; (b) a PlausibilityDeclarationHash computed as H(CanonicalConstraintArray || schema_id || schema_version); (c) a requirement that agents treat mirror pages as non-admissible upon encountering result FAIL with severity_level 4 or 5, absent an explicit agent-side policy override; and (d) a prohibition on disclosure of internal plausibility matrix logic, constraint weights, or evaluation algorithms through the declaration format.

  2. The format of claim 1 wherein constraint_class values may be extended via the schema extension framework.

  3. A computing system configured to generate plausibility declarations conforming to the format of any of claims 1 to 2.

  4. A non-transitory computer-readable medium storing instructions which, when executed, generate or validate plausibility declarations conforming to the format of any of claims 1 to 2.

Licence

Any person is granted a perpetual, irrevocable, worldwide, royalty-free licence to make, use, implement, modify, or distribute any system or method described in this disclosure for any purpose, without restriction, under the Apache License 2.0.

A copy of the Apache License 2.0 is available at https://www.apache.org/licenses/LICENSE-2.0

Abstract

A plausibility constraint declaration format is disclosed for the plausibility field of AIEP machine mirror pages. Each constraint entry comprises constraint_id, constraint_class, result (PASS/FAIL/WARN), and severity_level integer in [1, 5]. A PlausibilityDeclarationHash is computed as H(CanonicalConstraintArray || schema_id || schema_version). Agents must treat mirror pages as non-admissible upon encountering result FAIL with severity_level 4 or 5 absent an explicit policy override. Internal plausibility matrix logic is not disclosed through the format. The declaration enables structured agent consumption of structural coherence outcomes without disclosure of internal evaluation logic.

This specification is published as open-source prior art under the Apache License 2.0. Copyright 2026 Protocol Owner. Inventor: Protocol Architect.

Patent information · Licensing · All open specifications