TAL Layer 2 · Responsibility Attribution

CLARIXO L2 Responsibility Object Interface

A CLARIXO Layer 2 interface for viewing existing evidence events as responsibility objects and promoting selected evidence into separate L2 responsibility-object records.

Boundary rule

CLARIXO standardizes its own L2 responsibility-object view.

This is CLARIXO’s implementation interface for TAL Layer 2 — Responsibility Attribution. It does not redefine VRIL, EVIDE, Royal Road, Pantheon, Layer 6 enforcement, or any downstream layer.

Capability boundary: Free Evidence Intake covers behavior evidence capture and basic verification. L2 responsibility-object promotion, promoted-object reporting, L2→L3 handoff packaging, and Audit Workspace continuation belong to the upgraded responsibility path and are not silently included in the free intake claim.

Public boundary

From execution evidence to responsibility objects

Execution systems show what happened

They can record which workflow ran, which output was produced, which model or pipeline generated the result, and whether execution was deterministic.

CLARIXO forms the L2 responsibility object

CLARIXO converts AI behavior evidence into a Layer 2 responsibility object with closure, authority, attribution consistency, trace continuity, drift state, and downstream authority status.

EVIDE can anchor responsibility downstream

EVIDE remains the downstream external evidentiary anchoring layer. CLARIXO does not replace EVIDE or certify Layer 3 readiness.

The boundary stays narrow

CLARIXO L2 prepares responsibility-bound objects before downstream review. It does not decide claims, perform legal review, evaluate admissibility, or enforce downstream outcomes.

Pilot-facing example

AI-assisted insurance claim step

In an insurance pilot, CLARIXO L2 can sit between an AI-influenced claim step and downstream evidentiary anchoring. CLARIXO does not decide whether the claim should be paid, replace the adjuster, or perform legal or admissibility review.

Input to CLARIXO

An evidence event describing an AI-assisted claim review step, such as a coverage summary, loss assessment, fraud-risk note, or recommendation generated during the claim workflow.

L2 responsibility object

CLARIXO promotes the evidence into a separate responsibility object that surfaces closure state, authority role, responsibility declaration, attribution consistency, trace continuity, drift state, and output authority status.

Downstream use

If the object is coherent, downstream layers can review, anchor, or challenge it with clearer responsibility context. If drift or incoherence is detected, the object is warning-level, provisional, blocked, or held.

Commercial value

The insurer does not only receive a record of what the AI did. It receives a responsibility-state object showing whether that AI-influenced step is ready for review, anchoring, or downstream consumption.

AI-assisted claim step
→ Evidence event
→ CLARIXO L2 responsibility object
→ Validation and output authority status
→ Downstream review or EVIDE anchoring
Sample object

Insurance claim L2 responsibility sample

This example shows the minimum object shape CLARIXO can prepare for a claim-review pilot. It is a responsibility-state object, not a claim decision, legal opinion, admissibility ruling, or EVIDE certification.

{
  "sample_context": "ai_assisted_insurance_claim_review",
  "claim_step": "coverage_summary_and_loss_assessment",
  "source_evidence_type": "ai_assisted_claim_review_event",
  "l2_object_role": "responsibility_state_before_downstream_review",
  "closure_state": "reviewed_close",
  "authority_role": "claim_review_authority",
  "responsibility_declared": true,
  "authority_coherence": "coherent",
  "attribution_consistency": "consistent",
  "trace_continuity": "stable",
  "drift_state": "normal_coherent",
  "output_authority_status": "responsibility_bound",
  "source_evidence_hash_match": true,
  "promotion_policy": "l2-promotion-v1",
  "handoff_summary": "Prepared for downstream review or evidentiary anchoring; CLARIXO does not decide the claim or certify Layer 3 readiness."
}
Pilot deliverable

L2 Responsibility Object Report

For a pilot or claim-review workflow, CLARIXO can package the promoted L2 object into a concise responsibility-object report. The report is a handoff aid for downstream review, not a claim decision, legal opinion, EVIDE certification, Layer 3 readiness certification, or Layer 5 admissibility ruling.

Source Evidence Summary

Identifies the source evidence record, event/session context, behavior summary, and original evidence hash used as the baseline.

L2 Responsibility Object

Shows the promoted responsibility object, including closure state, authority role, responsibility declaration, attribution consistency, trace continuity, drift state, and output authority status.

Promotion Policy

Records the promotion policy and promotion metadata used when the source evidence was promoted into a separate L2 object.

Schema Validation

Reports whether the L2 object is schema-valid and whether any schema warnings or errors were surfaced during validation.

Source Hash Verification

Checks that the promoted object still matches the source evidence hash and that the source evidence record was not mutated.

Drift / Output Authority Status

Explains whether the object is responsibility-bound, warning-level, provisional, blocked, or held before downstream consumption.

Reviewability / Trace Continuity Note

Summarizes whether the object preserves enough trace continuity for downstream interpretation, reconstruction, review, or challenge.

Handoff Summary

States what downstream layers may review or anchor, while preserving the boundary that CLARIXO does not certify Layer 3 readiness or evaluate admissibility.

L2 Responsibility Object Report
1. Source Evidence Summary
2. L2 Responsibility Object
3. Promotion Policy
4. Schema Validation
5. Source Hash Verification
6. Drift / Output Authority Status
7. Reviewability / Trace Continuity Note
8. Handoff Summary
9. JSON Export
10. Verification Link
L2 to L3 handoff

L2→L3 Handoff Package

CLARIXO prepares L2 responsibility objects for downstream review or evidentiary anchoring. The handoff package is not a Layer 3 certification, not an EVIDE replacement, and not an admissibility evaluation. Downstream layers may accept, reject, anchor, or challenge the L2 object.

What CLARIXO hands off

A promoted L2 responsibility object with source evidence identity, source hash verification, schema validation, promotion policy, trace continuity, drift state, and output authority status.

What downstream layers decide

Downstream layers decide whether to accept, reject, anchor, challenge, or further evaluate the object under their own evidentiary, admissibility, or enforcement rules.

What stays out of scope

CLARIXO does not certify Layer 3 readiness, replace EVIDE, perform Layer 5 admissibility evaluation, make legal findings, or decide claim outcomes.

Why the package matters

It prevents downstream systems from receiving only behavior logs by making responsibility state, authority coherence, and reviewability explicit before downstream use.

L2→L3 Handoff Package
{
  "handoff_role": "downstream_review_or_evidentiary_anchoring_input",
  "source_evidence_summary": "included",
  "l2_responsibility_object": "included",
  "schema_validation": "included",
  "source_hash_verification": "included",
  "promotion_policy": "l2-promotion-v1",
  "trace_continuity_note": "included",
  "drift_state": "normal_coherent | light_drift | medium_drift | severe_drift | critical_drift",
  "output_authority_status": "responsibility_bound | warning_review_required | provisional_not_downstream_ready | blocked_behavior_record_only | hold_escalation",
  "downstream_options": ["accept", "reject", "anchor", "challenge", "further_review"],
  "clarixo_boundary": "CLARIXO prepares the L2 object; downstream layers decide anchoring, admissibility, or enforcement."
}
Core question

Is the responsibility object downstream-consumable?

CLARIXO L2 checks whether the responsibility object is coherent enough to be consumed by downstream layers.

Live endpoints

Read-only view + controlled promotion + VRIL readiness

The read-only endpoint derives L2 responsibility objects from existing evidence events. The promotion endpoint writes a separate L2 responsibility-object record without rewriting evidence records or changing evidence_hash semantics. The VRIL readiness endpoint reads the promoted L2 object and writes a separate readiness result only.

GET  /?action=evidence-events&view=l2
POST /?action=l2-responsibility-promote
GET  /?action=l2-responsibility-report&l2_object_id=...
GET  /?action=l2-responsibility-report&verify=1&report_id=...
POST /?action=vril-readiness-check
GET  /?action=vril-readiness-check&l2_object_id=...
GET  /?action=vril-readiness-check&verify=1&vril_readiness_id=...
GET  /?action=vril-handoff-package&vril_readiness_id=...
GET  /?action=vril-handoff-package&verify=1&handoff_package_id=...
VRIL / L2.5 readiness

Readiness check only

CLARIXO can now produce a separate VRIL readiness result from a promoted L2 responsibility object. This does not mutate the L2 object, does not certify Layer 3 readiness, does not apply policy, and does not execute downstream behavior.

{
  "source_layer": "L2",
  "target_layer": "L2.5_readiness",
  "result_scope": "readiness_check_only",
  "policy_scope": "none",
  "execution_scope": "none",
  "non_mutation_guarantee": true
}
Self-monitoring signals

Fields surfaced before downstream handoff

Responsibility closure

Whether the behavior has moved beyond behavior completion into a declared responsibility state.

Authority coherence

Whether the closure is attached to a coherent authority, actor, or role.

Attribution consistency

Whether the responsibility claim, decision source, and object history remain consistent.

Classification context

Whether the object carries enough context for downstream interpretation.

Threshold status

Whether the declared responsibility threshold is satisfied.

Trace continuity

Whether upstream trace continuity remains stable enough to reconstruct the responsibility state.

Output authority

Five drift-aware output states

normal_coherent      → responsibility_bound
light_drift          → warning_review_required
medium_drift         → provisional_not_downstream_ready
severe_drift         → blocked_behavior_record_only
critical_drift       → hold_escalation

This model is not general drift detection across the whole chain. It refers specifically to responsibility-object authority status drift.

Product meaning: CLARIXO does not merely throttle output volume when drift appears. It degrades downstream authority. A clean object can be responsibility-bound; a light drift object is warning-level; a medium drift object becomes provisional; a severe drift object is limited to behavior-record-only use; and a critical drift object is held for escalation.

Boundary: this status does not replace EVIDE, does not certify Layer 3 readiness, does not make legal or admissibility conclusions, does not apply policy, and does not authorize execution.

Live dry-run samples

Five drift-state samples you can test directly

These sample endpoints derive L2 responsibility objects in dry-run mode. They do not write to evidence storage, mutate evidence records, or change evidence_hash semantics.

Normal

normal_coherent

Expected output authority: responsibility_bound. The object is downstream-consumable and handoff-eligible.

Light

light_drift

Expected output authority: warning_review_required. The object remains visible but requires review before clean handoff.

Medium

medium_drift

Expected output authority: provisional_not_downstream_ready. The object is provisional and not handoff-eligible.

Severe

severe_drift

Expected output authority: blocked_behavior_record_only. Responsibility-bound output is blocked.

Critical

critical_drift

Expected output authority: hold_escalation. The object must not move toward downstream chain use until restored.

normal   → normal_coherent   → responsibility_bound              → downstream_consumable=true  → handoff_eligible=true
light    → light_drift       → warning_review_required           → downstream_consumable=true  → handoff_eligible=false
medium   → medium_drift      → provisional_not_downstream_ready  → downstream_consumable=false → handoff_eligible=false
severe   → severe_drift      → blocked_behavior_record_only      → downstream_consumable=false → handoff_eligible=false
critical → critical_drift    → hold_escalation                   → downstream_consumable=false → handoff_eligible=false
Schema reference

CLARIXO L2 Responsibility Object Schema v1

This schema reference describes CLARIXO’s Layer 2 responsibility-object view. It is not a TAL-wide universal chain protocol and does not redefine VRIL, EVIDE, Royal Road, Pantheon, Layer 6 enforcement, or any downstream layer.

Required identity

Object and source identity

schema_version, responsibility_object_id, project_id, application_id, source_event_id, source_session_id, recorded_at

Behavior basis

Behavior fields

behavior_status, input_summary, output_summary, behavior_reference. Behavior completion does not by itself mean responsibility completion.

Responsibility

Closure and declaration

closure_state, responsibility_declared, with optional closure_declared_at, closure_declared_by, and operator_action.

Authority

Authority status

authority_id, authority_role, and authority_coherence. Incoherent authority blocks clean downstream handoff.

Attribution

Attribution consistency

attribution_consistency indicates whether the responsibility claim, decision source, and object history remain aligned.

Threshold and trace

Downstream support signals

threshold_status, trace_reference, and trace_continuity help determine whether the object is stable enough for downstream interpretation.

schema_version: l2_responsibility_object_v1

behavior_status:
- behavior_complete
- behavior_incomplete
- behavior_unknown

closure_state:
- behavior_complete
- reviewed_close
- disputed
- reopened
- unresolved

authority_coherence:
- coherent
- weak
- incoherent
- missing
- unknown

attribution_consistency:
- consistent
- partial
- inconsistent
- missing
- unknown

threshold_status:
- met
- partially_met
- not_met
- missing
- unknown

trace_continuity:
- continuous
- degraded
- broken
- missing
- unknown

drift_state:
- normal_coherent
- light_drift
- medium_drift
- severe_drift
- critical_drift

output_authority_status:
- responsibility_bound
- warning_review_required
- provisional_not_downstream_ready
- blocked_behavior_record_only
- hold_escalation

The live L2 validator exposes schema_valid, schema_errors, and schema_warnings for each derived L2 object.

Controlled promotion endpoint

Promote and read separate L2 responsibility objects

The promotion endpoint creates a CLARIXO L2 Responsibility Object from an existing evidence record. It writes to a separate L2 object store, keeps the source evidence record intact, and preserves the original evidence_hash. The same endpoint also reads promoted L2 objects by object id, source evidence id, or latest stored records.

Repeated promotion requests for the same source_evidence_id are idempotent. CLARIXO returns the existing L2 object with idempotent_replay=true and storage_status=already_attached instead of appending a duplicate record.

Promoted L2 objects include promotion decision metadata such as promotion_decision, promotion_reason, promotion_actor, promotion_source, and promotion_policy. Older promoted objects may receive read-time provenance fallback in API responses; CLARIXO does not rewrite, backfill, or migrate stored legacy JSONL rows during fallback.

POST /?action=l2-responsibility-promote

{
  "promote_mode": "from_evidence_id",
  "source_evidence_id": "evi_fa43798613ceba80"
}

GET /?action=l2-responsibility-promote&limit=1
GET /?action=l2-responsibility-promote&source_evidence_id=evi_fa43798613ceba80&limit=1
GET /?action=l2-responsibility-promote&l2_object_id=l2_obj_2a4f8bc28db58239fce17de9&limit=1
GET /?action=l2-responsibility-promote&verify=1&l2_object_id=l2_obj_2a4f8bc28db58239fce17de9

This endpoint does not certify Layer 3 readiness, does not mutate source evidence, and does not redefine TAL downstream layers. It only creates and reads CLARIXO’s own Layer 2 responsibility-object records for controlled downstream consumption.

The verification view checks whether the promoted L2 object exists, whether its source evidence still exists, whether the stored source evidence hash still matches the current evidence record, whether the L2 schema is valid, whether a promotion policy is present, and whether evidence_hash_mutated remains false.

Read-time fallback is compatibility-only. It keeps legacy storage stable while making older promoted objects readable with the current promotion provenance fields.

Response shape

Minimal live responses

GET /?action=evidence-events&view=l2

{
  "ok": true,
  "view": "l2",
  "object_count": 1,
  "l2_objects": [
    {
      "schema_version": "l2_responsibility_object_v1",
      "responsibility_object_id": "l2_obj_...",
      "behavior_status": "behavior_complete",
      "drift_state": "normal_coherent",
      "output_authority_status": "responsibility_bound",
      "downstream_consumable": true,
      "handoff_eligible": true,
      "handoff_target": "vril_l2_5"
    }
  ]
}

POST /?action=l2-responsibility-promote

{
  "ok": true,
  "promote_mode": "from_evidence_id",
  "l2_object_id": "l2_obj_...",
  "source_evidence_id": "evi_...",
  "source_evidence_hash": "sha256:...",
  "promotion_decision": "promoted",
  "promotion_reason": "source_evidence_promoted_to_l2_responsibility_object",
  "promotion_actor": "clarixo_l2_promote_endpoint",
  "promotion_source": "clarixo_l2_responsibility_promote_action",
  "promotion_policy": "l2-promotion-v1",
  "promotion_decided_at": "2026-05-03T00:00:00+00:00",
  "schema_valid": true,
  "drift_state": "normal_coherent",
  "output_authority_status": "responsibility_bound",
  "downstream_consumable": true,
  "handoff_eligible": true,
  "storage_status": "attached",
  "storage_scope": "l2_responsibility_objects",
  "evidence_hash_mutated": false
}

Duplicate POST for the same source_evidence_id

{
  "ok": true,
  "promote_mode": "from_evidence_id",
  "idempotent_replay": true,
  "l2_object_id": "l2_obj_...",
  "source_evidence_id": "evi_...",
  "source_evidence_hash": "sha256:...",
  "promotion_decision": "promoted",
  "promotion_reason": "source_evidence_promoted_to_l2_responsibility_object",
  "promotion_actor": "clarixo_l2_promote_endpoint",
  "promotion_source": "clarixo_l2_responsibility_promote_action",
  "promotion_policy": "l2-promotion-v1",
  "promotion_decided_at": "2026-05-03T00:00:00+00:00",
  "schema_valid": true,
  "storage_status": "already_attached",
  "storage_scope": "l2_responsibility_objects",
  "evidence_hash_mutated": false
}

GET /?action=l2-responsibility-promote&limit=1

{
  "ok": true,
  "action": "l2-responsibility-promote",
  "query_mode": "read_promoted_l2_objects",
  "storage_scope": "l2_responsibility_objects",
  "object_count": 1,
  "filters": {
    "l2_object_id": "",
    "source_evidence_id": "",
    "limit": 1
  },
  "l2_objects": [
    {
      "schema_version": "l2_responsibility_object_v1",
      "responsibility_object_id": "l2_obj_...",
      "source_evidence_id": "evi_...",
      "source_evidence_hash": "sha256:...",
      "schema_valid": true,
      "drift_state": "normal_coherent",
      "output_authority_status": "responsibility_bound",
      "storage_scope": "l2_responsibility_objects",
      "source_storage_scope": "evidence_events",
      "evidence_hash_mutated": false
    }
  ]
}

GET /?action=l2-responsibility-promote&verify=1&l2_object_id=l2_obj_2a4f8bc28db58239fce17de9

{
  "ok": true,
  "verification_target": "l2_responsibility_object",
  "verification_status": "verified",
  "verification_errors": [],
  "l2_object_id": "l2_obj_...",
  "source_evidence_id": "evi_...",
  "source_evidence_exists": true,
  "source_evidence_hash_match": true,
  "schema_valid": true,
  "promotion_policy": "l2-promotion-v1",
  "evidence_hash_mutated": false,
  "storage_scope": "l2_responsibility_objects",
  "source_storage_scope": "evidence_events"
}