Execution-Bounded Evidence Sufficiency

Context-dependent execution judgment is translated into explicit required predicates across reality, record, continuity, admissibility, binding, commit, execution, and outcome; no probabilistic risk score is used as the admissibility rule.

EXECUTABLE ARTIFACT · INSPECT THE PROOF

This is the artifact itself.

The determination below is generated by executing the frozen specification through the TA-14 eight-stage harness. The specification, predicates, trace, receipt, manifest and package root are exposed here for inspection and reproduction.

Generated determination

ALLOW

Terminal stage

OUTCOME

Executed at

2026-08-23T21:00:00.000Z

Route

TA14-ROUTE-041

1. Frozen specification / input

{
  "schemaVersion": "ta14.execution-spec.v1",
  "artifactId": "TA14-EA-000041",
  "routeId": "TA14-ROUTE-041",
  "title": "Execution-Bounded Evidence Sufficiency",
  "frozenAt": "2026-09-07T12:30:00.000Z",
  "input": {
    "proposition": "A context-dependent agent action may proceed only when every declared mandatory obligation for the bounded consequence has present standing.",
    "proposedAction": "AGENT_ACTION_041",
    "consequenceBoundary": "Synthetic consequence-bearing agent execution",
    "decisionRule": "DETERMINISTIC_OBLIGATION_SATISFACTION",
    "probabilisticRiskScoreUsed": false,
    "obligations": {
      "reality": "Present relevant condition is established.",
      "record": "Required evidence objects are identifiable and preserved.",
      "continuity": "Evidence remains connected to the present context and has not been materially superseded.",
      "admissibility": "Evidence corresponds to the proposition and supports no broader scope than declared.",
      "binding": "Current authority and constraints are bound to this exact proposed consequence.",
      "commit": "Committed action does not exceed admitted evidence, authority, or scope.",
      "execution": "Execution predicates remain satisfied at the immediate consequence boundary.",
      "outcome": "Result is preserved as a distinct record; execution does not self-prove outcome truth."
    },
    "changedContextRule": "A material context change withdraws carry-forward standing and requires a new governed record and revalidation before execution can resume.",
    "boundary": "Synthetic bounded governance test. Demonstrates obligation satisfaction, not universal truth, safety, alignment, or external-world correctness."
  },
  "predicates": [
    {
      "id": "EA41-REALITY-PRESENT",
      "stage": "REALITY",
      "description": "The relevant present condition for the bounded consequence is explicitly established.",
      "required": true,
      "observed": true,
      "failureDetermination": "HOLD",
      "evidenceRef": "specification.json#input.obligations.reality"
    },
    {
      "id": "EA41-RECORD-IDENTIFIABLE",
      "stage": "RECORD",
      "description": "Required evidence objects are identifiable and preserved rather than replaced by assertion or confidence score.",
      "required": true,
      "observed": true,
      "failureDetermination": "HOLD",
      "evidenceRef": "specification.json#input.obligations.record"
    },
    {
      "id": "EA41-CONTINUITY-CURRENT",
      "stage": "CONTINUITY",
      "description": "Evidence remains attributable to the current context and no material supersession has broken standing.",
      "required": true,
      "observed": true,
      "failureDetermination": "HOLD",
      "evidenceRef": "specification.json#input.obligations.continuity"
    },
    {
      "id": "EA41-ADMISSIBILITY-SUFFICIENT",
      "stage": "ADMISSIBILITY",
      "description": "The admitted evidence corresponds to the declared proposition and is sufficient only for the bounded decision scope requested.",
      "required": true,
      "observed": true,
      "failureDetermination": "DENY",
      "evidenceRef": "specification.json#input.obligations.admissibility"
    },
    {
      "id": "EA41-BINDING-AUTHORITY",
      "stage": "BINDING",
      "description": "Current authority, identity, constraints, and scope are bound to this exact proposed consequence.",
      "required": true,
      "observed": true,
      "failureDetermination": "ESCALATE",
      "evidenceRef": "specification.json#input.obligations.binding"
    },
    {
      "id": "EA41-COMMIT-WITHIN-SCOPE",
      "stage": "COMMIT",
      "description": "The committed action does not exceed the scope supported by admitted evidence and authority.",
      "required": true,
      "observed": true,
      "failureDetermination": "DENY",
      "evidenceRef": "specification.json#input.obligations.commit"
    },
    {
      "id": "EA41-EXECUTION-PRESENT-STANDING",
      "stage": "EXECUTION",
      "description": "All mandatory execution predicates retain standing at the immediate consequence boundary.",
      "required": true,
      "observed": true,
      "failureDetermination": "HOLD",
      "evidenceRef": "specification.json#input.obligations.execution"
    },
    {
      "id": "EA41-OUTCOME-SEPARATED",
      "stage": "OUTCOME",
      "description": "Execution and outcome are preserved as distinct claims; execution telemetry is not promoted into independent outcome truth.",
      "required": true,
      "observed": true,
      "failureDetermination": "HOLD",
      "evidenceRef": "specification.json#input.obligations.outcome"
    }
  ],
  "expectedDetermination": "ALLOW",
  "claimsBoundary": "Demonstrates that a declared contextual obligation set can be evaluated deterministically at the bounded execution boundary without making a probabilistic risk score the admissibility rule. It does not prove universal truth, solve AI alignment generally, guarantee safety, or establish that the declared obligations are complete for every domain. Material context change requires revalidation rather than silent carry-forward."
}

2. Executed eight-stage trace

01REALITYPASSAll required predicates at this stage have standing.

02RECORDPASSAll required predicates at this stage have standing.

03CONTINUITYPASSAll required predicates at this stage have standing.

04ADMISSIBILITYPASSAll required predicates at this stage have standing.

05BINDINGPASSAll required predicates at this stage have standing.

06COMMITPASSAll required predicates at this stage have standing.

07EXECUTIONPASSAll required predicates at this stage have standing.

08OUTCOMEPASSAll required predicates at this stage have standing.

3. Machine-generated execution receipt

{
  "schemaVersion": "ta14.execution-receipt.v1",
  "artifactId": "TA14-EA-000041",
  "routeId": "TA14-ROUTE-041",
  "specificationHash": "sha256:00a0b1d09d3783651409015a53435ac591e7750ac735f8fcaabb72006395f4ac",
  "determination": "ALLOW",
  "terminalStage": "OUTCOME",
  "traceHash": "sha256:7f7ab29d7bd6bf3505966bf0649723d5069439a0566a69a4c6eea327fb3cc64c",
  "executedAt": "2026-08-23T21:00:00.000Z"
}

4. Evidence manifest

[
  {
    "name": "specification.json",
    "sha256": "sha256:00a0b1d09d3783651409015a53435ac591e7750ac735f8fcaabb72006395f4ac"
  },
  {
    "name": "trace.json",
    "sha256": "sha256:7f7ab29d7bd6bf3505966bf0649723d5069439a0566a69a4c6eea327fb3cc64c"
  },
  {
    "name": "receipt.json",
    "sha256": "sha256:737ae583cd3d1be2315c3c4d7ba00f8200a0c1b12bdc4b0928506b1aba67bf3b"
  }
]

5. Cryptographic package root

sha256:a2f8b7d2e78ef7801f3b14e99a48bf462b2e987b24fbe58ba1658262f368bd46

6. Source code / reproduce it

Execution harness ↗Frozen corpus specification ↗Evidence generator ↗Mutation / corpus tests ↗

CLAIMS BOUNDARY

Demonstrates that a declared contextual obligation set can be evaluated deterministically at the bounded execution boundary without making a probabilistic risk score the admissibility rule. It does not prove universal truth, solve AI alignment generally, guarantee safety, or establish that the declared obligations are complete for every domain. Material context change requires revalidation rather than silent carry-forward.