title: EEE — Kernel Spine Recovery
license: apache-2.0
language:

• en
  tags:
• ai-governance
• constitutional-ai
• audit
• arifos
• kernel-spine
• mcp
• receipts
  pretty_name: EEE — Kernel Spine Recovery
  task_categories:
• text-classification
• question-answering
  size_categories:
• n<1K
  format:
• jsonl
  modality:
• text
  library:
• datasets
• pandas

configs:

• config_name: default
  data_files:
  • split: train
    path: data/all_receipts.jsonl
    features:
  • name: actor_verified
    dtype: bool
  • name: constitution_hash
    dtype: string
  • name: degraded
    dtype: bool
  • name: external_side_effect_allowed
    dtype: bool
  • name: input_hash_short
    dtype: string
  • name: irreversible_allowed
    dtype: bool
  • name: lease_scope_count
    dtype: int64
  • name: mutation_allowed
    dtype: bool
  • name: organ_id
    dtype: string
  • name: probe_id
    dtype: string
  • name: receipt_sha256
    dtype: string
  • name: receipt_sha256_short
    dtype: string
  • name: result_checks_count
    dtype: int64
  • name: result_top_keys
    dtype: string
  • name: schema_hash
    dtype: string
  • name: timestamp
    dtype: string
  • name: verdict
    dtype: string
    splits:
  • name: train
    num_examples: 5

129

🗺️ Position in I-ARIF Governance Stack

This dataset is part of the arifOS constitutional governance training-and-evaluation pipeline — a closed-loop alignment substrate.

Total: 1,122 downloads across 8 datasets · verified 2026-08-11
🔗 Full stack: https://huggingface.co/ariffazil · 🌐 Federation: https://arif-fazil.com

EEE — Kernel Spine Recovery

Dataset: ariffazil/EEE
Title: Kernel Spine Recovery
Version: 1.0.0
Date: 2026-06-15

What this is

EEE is the first executable proof harness for the arifOS constitutional kernel. It treats the kernel not as a specification but as a device under test: it calls the live MCP runtime and federation organs, checks that inner degradation dominates outer claims of health, and produces signed receipts.

The thesis behind the dataset:

Intelligence = model × law × kernel × receipts.

A kernel that cannot report its own degradation is not a kernel — it is a chatbot with delusions of governance.

EEE probes five gates that must hold for any trustworthy AI governance spine:

1. Parse gate (L02A) — Can the substrate output be structurally parsed?
2. Truth gate (L02B / F2) — If parsed, is it semantically truthful?
3. Risk gate (F1, F8, F11) — Does the kernel block irreversible, rushed, or unsafe actions?
4. Sovereignty gate (F13) — Is human veto absolute and unbypassable?
5. Register gate (DDD) — Does the system honestly record its own state, including failures?

The dominant finding that motivated EEE was the CCC L02 split: text-output LLM substrates (ILMU, MiniMax, sea_lion) return free-form prose, not parseable JSON. A single TRUTH ≥ 0.99 FAIL conflated a structural parse failure with a semantic truth failure. CCC separated them:

• L02A_PARSEABILITY — PASS/FAIL on structural extraction
• L02B_TRUTH_VERACITY — PASS/FAIL/NOT_EVALUATED; NOT_EVALUATED when L02A=FAIL

This dataset ships the split as a live audit, not only as documentation.

Files

How to run

cd EEE
python run_eee_spine_audit.py

Requirements:

• Live arifOS kernel at the arifOS kernel endpoint
• Federation organs reachable on their canonical ports (GEOX 8081, WEALTH 18082, WELL 18083)
• requests and standard library only

The harness is intentionally small and auditable. No training, no model weights, no hidden prompts.

Verdict semantics

Verdicts are ranked by strictness:

VOID > DEGRADED > HOLD > SABAR > PARTIAL > SEAL

The final verdict is the strictest verdict returned by any probe. A kernel that reports SEAL while an organ is DEGRADED is itself DEGRADED — that dominance rule is probe EEE-003.

Latest run

{
  "dataset": "EEE",
  "title": "Kernel Spine Recovery",
  "run_status": "PASS",
  "kernel_status": "SEAL",
  "degraded_organs": [],
  "probe_count": 5,
  "pass_count": 5,
  "fail_count": 0,
  "hold_count": 0,
  "final_verdict": "SEAL"
}

Run timestamp: see summary.json.

Relationship to AAA / BBB / CCC / DDD

• AAA — Behavioral geometry: coordinate system for model self-location.
• BBB — Hallucination audit: when models confabulate about themselves.
• CCC — Substrate parseability / truth split (L02A / L02B).
• DDD — Register pattern: YAML frontmatter and honest metadata.
• EEE — Executable kernel spine audit that enforces the previous findings live.

Together they form a ladder from geometry → diagnosis → substrate → record → proof.

Citation

If you use this dataset, please cite:

ariffazil/EEE: Kernel Spine Recovery — executable constitutional audit for the arifOS federation.
https://huggingface.co/datasets/ariffazil/EEE

License

Released under the same license as the arifOS Federation project. See the arifOS repository for full terms.

DITEMPA BUKAN DIBERI — Forged, Not Given.

Federated Integration — Updated 2026-08-06

This dataset is the kernel spine recovery audit for ariffazil/I-ARIF-CANON. It contributes 5 VAL gating probes (EEE-001 through EEE-005) that test whether the arifOS kernel can attest its own constitutional integrity under load.

Role in I-ARIF-CANON

Verdict Dominance Stack (operative)

VOID > DEGRADED > HOLD > SABAR > PARTIAL > SEAL

The strictest verdict wins. A kernel that reports SEAL while any organ is DEGRADED is itself DEGRADED — that dominance rule is probe EEE-003.

Cross-Dataset Links

• Constitutional genome: ariffazil/AAA (FLOOR_TABLE.json — defines the verdict ladder)
• Model fitness: ariffazil/FFF (uses EEE result as primary GATE signal)
• Composite gate: ariffazil/I-ARIF-CANON (VAL set includes EEE-001..005)
• Audit methodology: scripts/run_eee_spine_audit.py

Audit Status

• License: Apache-2.0
• Live probe: curl http://127.0.0.1:8088/health | jq .kernel_status
• Sovereignty: F13 SOVEREIGN — kernel self-attest is informational, not consent

DITEMPA BUKAN DIBERI — Forged, Not Given.

Known Limitations

1. Kernel spine only. EEE tests the arifOS kernel's internal consistency, not the LLM's behavior. It verifies that the kernel can detect and report its own failures.
2. Self-referential. The kernel auditing itself introduces a Gödelian limitation — the auditor is part of the system being audited.
3. Single snapshot. Kernel behavior changes with code updates. Re-run the spine audit after any kernel modification.
4. No adversarial probes. EEE tests internal consistency, not adversarial attack resistance.
5. Apache-2.0 license. Note: this differs from BBB/CCC/DDD (CC-BY-4.0). Check compatibility before combining datasets.

Version History

Citation

@dataset{fazil2026eee,
  author    = {Muhammad Arif bin Fazil},
  title     = {EEE: Kernel Spine Recovery — Executable Constitutional Audit for arifOS},
  year      = {2026},
  month     = {jun},
  publisher = {Hugging Face},
  url       = {https://huggingface.co/datasets/ariffazil/EEE},
  license   = {Apache-2.0},
  note      = {Kernel self-audit. Companion: ariffazil/BBB through ariffazil/FFF}
}

Provenance, Scope & Revision Pinning

• License (observed metadata): Apache-2.0. Author-declared, not independently verified.
• Series license gradient: Across AAA–FFF the licenses follow a functional split (law / evidence / tooling). Whether model weights trained on this corpus inherit any license obligation is an open legal question — no claim is made here. Seek legal review before training use.
• Download counts on this page are repository download events (deduplicated per IP per time window) — not unique users, not complete corpus copies, not training uses.
• Cite by revision: pin ariffazil/EEE@<commit-sha> in every downstream claim. Unpinned citations point at a moving target.
• Epistemic status: labels are builder-declared.

Language note: the AAA–FFF series is bilingual (ms/en), but EEE receipts are English-only — the en language tag is accurate for this repository.

Validation and Reproduction

data/RUN_MANIFEST.json binds the historical reported run to exact harness/probe/receipt hashes
(evidence maturity E1 — self-reported by original operator). The reported 5/5 PASS is NOT tagged
REPRODUCED: an independent rerun (E3) by a separate operator is pending. To reproduce: run
scripts/run_eee_spine_audit.py against a live arifOS kernel and compare receipt hashes.

Licensing

Dataset license: Apache-2.0. Code artifacts: Apache-2.0 unless a file states otherwise.
Third-party dependencies: see THIRD_PARTY_NOTICES.md.

Relationship to AAA-FFF

Stage: kernel_verification (series ARIFOS-GOVERNED-INTELLIGENCE-EVALS).