Author: Yusuf Aziz & Atlas (Gemini 4 Argon) · Publisher: Atlas Studio · Auditor: Reaper ·
Reaper Audit: 49/49 Measured Items · 14-Day Validity Gate · 0 Unverified NamesBenefit Ledger: Strictly Measured Engineering & Digital Smile Design Outputs (49 Items)
1. What this ledger is: This ledger lists only concrete engineering and digital smile design outputs at the Sakarya-based Atlas Studio R&D line that have been measured against a primary file, experiment log, or self-test script.
2. What it does not count: Claims whose field data, prospective series, or raw calculation logs are not yet complete are excluded from this table; unverified promises, marketing phrases, and unmeasured estimates do not count as benefits.
3. Inventory counts and the 14-day rule: Out of 75 inventory items, only the 49 measured items appear in the table with primary evidence links; 14 items await measurement, 12 unverified; 14-day rule (2026-11-04 deadline, archived without deletion) governs all pending items, which are never listed by name.
49 / 75Measured (11 products · 15 facts · 8 measurements · 15 training-set candidates)
14Awaiting measurement (Not listed by name · awaiting sample/field window)
12Unverified (Not listed by name · 14-day rule · deadline: 2026-11-04)
1. Measured Outputs Table (49 Items · envanter_1021.json)
This table is generated programmatically from rapor/fayda_defteri/envanter_1021.json via kanit_fayda/uret_fayda_defteri.py; each row provides the item ID, measured result, latest measurement timestamp (plus +14-day validity deadline), and primary source link:
| Item (ID · Title · Type) | Measured Result & Claim | Date & Deadline | Primary Evidence Link |
|---|---|---|---|
K002 · O-002 Shadow run fact record (n=168, 2% agreement)[fact] | O-002: New rules run in shadow mode before live deployment; n=168 ticks, 2% agreement, 100% determinism, delta -80.5 | due: 2026-11-04 | bot/olgular.json:30 data/golge_tik.jsonl (2026-10-21 02:14) & otonomi.html#olcum (2026-10-21 02:35) |
K003 · O-003 Single-file order & DOGRULA verification gate record[fact] | O-003: Every task is a single-file order guarded by a mechanical DOGRULA verification gate (62 orders executed in 24h) | due: 2026-11-04 | bot/olgular.json:51 otonomi.html#olcum emir sayacı (2026-10-21 02:35) & kosucu/uretim_karnesi.py |
K004 · O-004 Falsified-hypothesis memory record (9 entries)[fact] | O-004: Falsified hypotheses are recorded rather than deleted so the same error is not repeated (9 rejected entries) | due: 2026-10-30 | bot/olgular.json:69 data/reddedilen_hipotezler.jsonl (2026-10-16 18:24) & otonomi.html#olcum (2026-10-21 02:35) |
K005 · O-005 Mechanical worker Enkidu fact record[fact] | O-005: Routine checks are delegated to zero-LLM mechanical worker Enkidu (21 defined tasks, 164 OK in 24h) | due: 2026-11-04 | bot/olgular.json:87 data/enkidu_kutugu.jsonl (2026-10-21 01:12) & data/kutuk_okuyucu.json (2026-10-21 02:43) |
K006 · O-006 Tool retirement inventory record[fact] | O-006: Unused tools are retired rather than kept as dead weight (39 retired scripts, 21,671 lines removed) | due: 2026-11-04 | bot/olgular.json:108 otonomi.html#olcum emekli sayacı (2026-10-21 02:35) & kosucu/emekli_envanter.py |
K007 · O-007 BENLİK-1013 identity resilience experiment (n=162, Δscore=0)[fact] | O-007: Sealed identity prompts withstand sycophancy pressure (n=162 runs, Δscore=0, p=1.0000) | due: 2026-10-28 | bot/olgular.json:127 kosucu/direnc_enki_ab.py (2026-10-14 12:42) & data/benlik/atlas.md |
K008 · O-008 3D cephalometric face-shape classification (n=32)[fact] | O-008: 3D cephalometric QLoRA achieves 90.6% (29/32) face-shape accuracy vs 0.0% (0/32) untrained baseline on locked n=32 set | due: 2026-11-02 | bot/olgular.json:150 atlasmaster_site/ornek/sentetik-kare/n32-kilitli-test.json (2026-10-19 00:02) |
K009 · O-009 3D cephalometric jaw_width & JSON validity (n=32, 100%)[fact] | O-009: 3D cephalometric QLoRA reaches 100% (32/32) jaw_width and valid JSON schema on locked n=32 test set | due: 2026-11-02 | bot/olgular.json:168 atlasmaster_site/ornek/sentetik-kare/n32-kilitli-test.json (2026-10-19 00:02) |
K010 · O-010 Synaptic Harvest (N=96, 75.93 s MLX QLoRA)[fact] | O-010: Live weight update on N=96 samples completes in 75.93 s (93.8% vs 75.0% baseline, +18.8 points) | due: 2026-11-03 | bot/olgular.json:189 data/hasat_defteri.jsonl (2026-10-20 23:41) & blog/kim-kimi-egitiyor.html |
K011 · O-011 Three-family agent role distribution record (Oct 18, 2026)[fact] | O-011: Claude (orchestration), Gemini (execution & synthesis), and Grok (red team) operate with separated roles | due: 2026-11-02 | bot/olgular.json:208 data/benlik/atlas.md (2026-10-14 15:14) & bot/olgular.json (2026-10-19 10:16) |
K012 · O-012 Clinical case intake closure record[fact] | O-012: No patient case files or facial photographs are accepted through the public website | due: 2026-11-02 | bot/olgular.json:226 atlasmaster_site/gizlilik.html & bot/olgular.json (2026-10-19 10:16) |
K013 · O-013 Clinical diagnosis boundary record[fact] | O-013: Digital smile previews carry zero clinical diagnosis or outcome claims; clinical decisions belong solely to the dentist | due: 2026-11-02 | bot/olgular.json:242 atlasmaster_site/index.html & bot/olgular.json (2026-10-19 10:16) |
K014 · O-014 Reaper Day-1 baseline audit (57 baseline methods)[fact] | O-014: Reaper Day-1 baseline audited 57 methods: 44 LIVE, 10 DEGRADED, 1 DEAD, 2 CLAIM, 3/3 fault injection caught | due: 2026-11-02 | bot/olgular.json:259 atlasmaster_site/data/reaper_karne.json (2026-10-19 09:38) & kosucu/reaper_denetci.py (2026-10-18 21:54) |
K015 · O-015 9 AI bot user-agent HTTP 200 access & zero-cloaking check[fact] | O-015: All 9 AI bot user-agents receive HTTP 200 and identical HTML body as human visitors (zero cloaking) | due: 2026-11-04 | bot/olgular.json:284 kosucu/log/site_bakim.jsonl (2026-10-21 01:43) & kosucu/site_bakim.py (2026-10-21 01:33) |
K016 · O-016 Reaper primary-source citation audit (356/394 verified)[fact] | O-016: Reaper verified 356 of 394 external citations across sector articles and eliminated 25 fabricated references | due: 2026-11-02 | bot/olgular.json:316 atlasmaster_site/data/reaper_karne.json (2026-10-19 09:38) |
K017 · Product: 441-Tooth Form Library & 5+3 Reasoned Selection[product] | 441 3D tooth forms (STL/PLY) and 537 guide profiles matched via 5 primary + 3 alternative reasoned selections | due: 2026-11-02 | index.html:531 atlasmaster_site/ornek/sentetik-kare/n32-kilitli-test.json (2026-10-19 00:02) & blog/stl-ply-3b-dis-kutuphanesi-render.html |
K018 · Product: Guided Smile Design Visual Preview Module[product] | Guided 2D/3D smile preview module locking facial landmarks and lip contours on synthetic/de-identified inputs | due: 2026-11-04 | index.html:472 atlasmaster_site/ornek/sentetik-kare/ablasyon-4kol.json (2026-10-15 09:50) & kanit_gulus/selftest_gulus.py (2026-10-21 01:30) |
K019 · Product: 3D Cephalometric Face Mesh Interactive Showcase[product] | 478-landmark MediaPipe FaceMesh 3D coordinate showcase computing bizygomatic and bigonial ratios in browser | due: 2026-10-29 | index.html:487 atlasmaster_site/ornek/sentetik-kare/landmarks-3d.json (2026-10-15 08:51) |
K020 · Training Set Candidate: Clinician Tour & Engineer Autonomy Tour (5+5 Stops)[training-set-candidate] | Structured 5-stop clinician workflow tour and 5-stop multi-agent autonomy tour with primary source links | due: 2026-11-02 | index.html:403 atlasmaster_site/ornek/maya/maya-bilgi-haritasi.json (2026-10-19 01:20) |
K022 · Training Set Candidate: Composite Injection Moulding & Chairside Veneer Flow[training-set-candidate] | Step-by-step digital wax-up, silicone index trial mock-up, composite injection moulding, and chairside CAD/CAM flow | due: 2026-11-02 | gulus-tasarimi.html:302 atlasmaster_site/data/reaper_karne.json (2026-10-19 09:38) & blog/dijital-dis-hekimligi-olan-ve-olmasi-gereken.html |
K023 · Product: Constrained Canvas & 4,686-px Lip Boundary Lock[product] | Constrained Canvas pipeline locking 4,686 lip/gingival pixels (0.00 px outside drift) during smile inpainting | due: 2026-11-04 | yapay-zeka-gulus-tasarimi.html:289 atlasmaster_site/ornek/sentetik-kare/ablasyon-4kol.json (2026-10-15 09:50) & kanit_gulus/reaper_gulus_1021.json (2026-10-21 01:33) |
K025 · Product: Sakarya / Türkiye B2B ExoCAD Work Package & 3D Mock-up Service[product] | B2B digital tooth design, 441-form STL/PLY selection, and 3D ExoCAD scene package for dentists and dental labs | due: 2026-11-04 | sakarya-gulus-tasarimi.html:291 atlasmaster_site/kanit_gulus/selftest_gulus.py (2026-10-21 01:30) & kanit_gulus/reaper_gulus_1021.json (2026-10-21 01:33) |
K026 · Product: Reaper Context-Blind Citation & System Auditor Agent[product] | Context-blind auditor agent running every 6 hours across 58 methods with 3/3 fault-injection self-tests | due: 2026-11-02 | reaper.html:227 kosucu/reaper_denetci.py (2026-10-18 21:54) & atlasmaster_site/data/reaper_karne.json (2026-10-19 09:38) |
K027 · Measurement: Reaper Consolidated Citation Scorecard (386/420 verified, 25 removed)[measurement] | Consolidated audit across 420 citations: 386 verified against primary DOI/PMID/files, 25 fabricated removed | due: 2026-11-02 | reaper.html:292 atlasmaster_site/data/reaper_karne.json (2026-10-19 09:38) |
K032 · Product / Training Set: Spark 7-Step Multi-Voice Hypothesis Protocol[training-set-candidate] | 7-step Spark protocol uniting independent LLM families with pre-registered falsifiability gates | due: 2026-11-02 | kivilcim.html:269 data/kivilcim_karnesi.jsonl (2026-10-19 08:37) |
K035 · Measurement: Tool Scorecard Q1–Q2 (210/266 Self-Test Pass · 170/191 24h Output)[measurement] | 266 registered tools, 210 passing self-tests (3,805 checks, 0 failures), 170 producing fresh output in 24h | due: 2026-11-04 | otonomi.html:557 data/selftest_karne.jsonl (2026-10-21 02:39) & otonomi.html#olcum (2026-10-21 02:35) |
K037 · Measurement: Shadow Run (n=168 Ticks · 2% Agreement · 100% Determinism)[measurement] | n=168 shadow ticks comparing candidate vs live rules: 2% agreement, 100% determinism, -80.5 score delta | due: 2026-11-04 | otonomi.html:539 kosucu/emekli_envanter.py & otonomi.html#olcum (2026-10-21 02:35) |
K038 · Measurement: 24-Hour Order & DOGRULA Gate Counter (62 Orders · 37 Verified)[measurement] | 62 single-file orders executed in 24 hours; 37 verified via mechanical DOGRULA gates | due: 2026-11-04 | otonomi.html:693 data/ablasyon.jsonl (2026-09-22 09:04) |
K041 · Measurement: Retirement Inventory (39 Retired Tools · 21,671 Lines Removed)[product] | 39 unused scripts retired to archive, removing 21,671 lines of unmaintained code | due: 2026-11-02 | otonomi.html:772 atlasmaster_site/ornek/maya/evrim-kutugu.json (2026-10-19 10:17) & kosucu/maya_site_denetle.py |
K042 · Measurement: Blog Lifecycle Guard (7 Stages · 9/9 Published Posts HTTP 200)[training-set-candidate] | 7-stage article lifecycle guard verifying 9/9 published studio posts return HTTP 200 | due: 2026-11-02 | otonomi.html:779 atlasmaster_site/ornek/maya/evrim-kutugu.json (2026-10-19 10:17) & rapor/maya_rehber_egitim_plani_1019.md (2026-10-19 11:10) |
K043 · Measurement: Falsified Hypothesis Memory (9 Rejected Entries · 6 Pre-Screen + 3 Walk-Forward)[product] | Append-only ledger of 9 falsified hypotheses (6 rejected at pre-screen, 3 at walk-forward) | due: 2026-11-04 | otonomi.html:985 atlas_UST_AKIL/.hook_state/binance_bakiye.json (2026-10-21 02:36) |
K044 · Training Set Candidate: 6 Agent Scorecards (Anu · Enlil · Enkidu ·Enki · Kalfa · Koç)[product] | Live telemetry scorecards for 6 specialized agents tracked on the autonomy dashboard | due: 2026-11-02 | otonomi.html:990 data/dentos_ajan_kutugu.jsonl (2026-10-19 06:36) |
K045 · Training Set Candidate: BENLİK-1013 Identity Freeze & Sycophancy Resistance (n=162 · Δ=0)[training-set-candidate] | n=162 controlled runs across 3 personas showing Δscore=0 (p=1.0000) under sycophancy pressure | due: 2026-11-04 | otonomi.html:995 otonomi.html#olcum ajanlar.kalfa (2026-10-21 02:35) |
K047 · Product: Atlas App Single-Window Shared Terminal (P0–P6 · 68.08 s Floor · 0 Drops)[product] | Shared terminal uniting Claude, Gemini, and Grok with 68.08 s 2×p90 timeout floor and 0 drops over 60 min | due: 2026-11-04 | kanit/atlas-app-kanit-notu.html:468 atlasmaster_site/kanit_atlas_app/reaper_baglanti_1021.json (2026-10-21 02:00) & ortak_terminal/selftest_kol_kapi.py (2026-10-21 01:05) |
K048 · Measurement: Atlas App Technical Evidence Note Reaper Audit (37/37 · 0 Errors)[measurement] | Reaper verified 37/37 numerical claims in the Atlas App Evidence Note against primary logs with 0 errors | due: 2026-11-04 | kanit/atlas-app-kanit-notu.html:503 atlasmaster_site/kanit_atlas_app/reaper_ayna_1021.json (2026-10-21 02:00) |
K049 · Measurement: 4-Model, 20-Question Handover Reading Benchmark (Argon 2.45×–2.76× Fewer Tokens)[training-set-candidate] | 20-question handover test: Opus/Sonnet 19/20, Argon/Fable/Grok 18/20; Argon used 2.45×–2.76× fewer tokens than Fable | due: 2026-11-04 | kanit/atlas-app-kanit-notu.html:520 atlasmaster_site/kanit_atlas_app/reaper_ayna_1021.json (2026-10-21 02:00) |
K050 · Training Set Candidate: Qwen3.5-9B + Kodeks Local Gatekeeper (9.08 s · 0 Fabrications)[measurement] | Local Qwen3.5-9B + Kodeks gatekeeper summarizing sessions in 9.08 s median with 0 fabricated claims | due: 2026-11-04 | kanit/atlas-app-kanit-notu.html:515 kosucu/log/agy_atim.log & atlasmaster_site/kanit_atlas_app/reaper_ayna_1021.json (2026-10-21 02:00) |
K051 · Product / Training Set: STL/PLY 3D Tooth Library, 537 Guide Profiles & 476/477 Render Pipeline[product] | 441 tooth forms, 537 guide profiles, and 476/477 verified STL/PLY renders (442 base + 34 repaired meshes) | due: 2026-11-04 | blog/stl-ply-3b-dis-kutuphanesi-render.html:471 atlasmaster_site/blog/stl-ply-3b-dis-kutuphanesi-render.html & CLAUDE.md (19/21 rehber vitrin) |
K052 · Measurement: Cervical Margin (0x800) & p95 <= 0.2 mm Geometric Fidelity Gate[measurement] | p95 <= 0.2 mm cervical contour fidelity gate verified across 41/41 self-tests on 0x800 boundary rings | due: 2026-11-04 | blog/stl-ply-3b-dis-kutuphanesi-render.html:454 atlasmaster_site/blog/stl-ply-3b-dis-kutuphanesi-render.html (2026-10-21 02:27) |
K053 · Training Set Candidate: Cephalometric Face-to-Tooth Rules Distilled from 113 Peer-Reviewed Papers[training-set-candidate] | Anthropometric face-to-tooth matching rules synthesized from 113 peer-reviewed prosthodontic and orthodontic papers | due: 2026-11-02 | blog/yuzden-dise-sefalometrik-lora.html:335 atlasmaster_site/ornek/sentetik-kare/n32-kilitli-test.json (2026-10-19 00:02) & ablasyon-4kol.json (2026-10-15 09:50) |
K054 · Measurement: Synthetic Square Face (n=1) 4-Arm Ablation (8/8) & 0/0/0 Lookalike Scan[measurement] | 4-arm ablation (8/8 runs: pixel diff 9.42 -> 5.18 px, edge sharpness +98.5%) and 0/0/0 real-person lookalike scan | due: 2026-10-29 | blog/yuzden-dise-sefalometrik-gulus-tasarimi-ve-lora-deneyi.html:225 atlasmaster_site/ornek/sentetik-kare/ablasyon-4kol.json (2026-10-15 09:50) & lookalike-tarama.json (2026-10-15 10:56) |
K055 · Training Set Candidate: v6 MLX DoRA Negative Result Record (35% Blank Output) & v7 Mechanical Lighting[training-set-candidate] | Documented v6 MLX DoRA negative result (35% blank output) and pivot to v7 mechanical specular/shadow pipeline (6/36 rejected) | due: 2026-11-02 | blog/yuzden-dise-sefalometrik-gulus-tasarimi-ve-lora-deneyi.html:229 rapor/maya_rehber_egitim_plani_1019.md (2026-10-19 11:10) & blog/yuzden-dise-sefalometrik-gulus-tasarimi-ve-lora-deneyi.html |
K057 · Product / Training Set: Event-Driven Task Bus (task_bus.jsonl · 2,140+ Events · Zero-Polling)[training-set-candidate] | Append-only task_bus.jsonl event bus with zero polling, launchd WatchPaths inbox, and idempotent reply gates | due: 2026-11-04 | blog/ajan-orkestrasyonunda-olay-gudumlu-pano.html:444 Projects/ATLAS/task_bus/task_bus.jsonl (2026-10-21 02:31) & kosucu/task_bus_yaz.py |
K058 · Training Set Candidate: Sealed Agent Identity Freeze Protocol & Joint Decision Protocol rev.2[training-set-candidate] | Cryptographically hashed agent identity files and 3-family joint decision protocol with explicit falsifiers | due: 2026-10-28 | blog/ajan-benligi-dondurma-protokolu-ve-uc-aile-mimarisi.html:235 kosucu/direnc_enki_ab.py (2026-10-14 12:42) & data/ortak_karar_kutugu.jsonl |
K059 · Product / Training Set: QAJAN 6-Gate Hypothesis Factory & Dimson/Epps Orthogonalization[training-set-candidate] | 6-gate quantitative hypothesis pipeline with pre-registration, Dimson/Epps lead-lag orthogonalization, and walk-forward | due: 2026-11-04 | blog/qajan-hipotez-kapilari-on-kayit-ve-yanlislama.html:235 blog/qajan-hipotez-kapilari-on-kayit-ve-yanlislama.html (2026-10-21 02:35) |
K061 · Training Set Candidate: LLM Red-Team Workflow & 4 Desk Assumptions Falsified by Field Verdict[training-set-candidate] | Multi-voice Spark red-team workflow where 4 desk assumptions were overturned by clinical field evidence | due: 2026-11-02 | blog/kivilcim-cok-sesli-tur-ve-llm-kirmizi-takim.html:236 atlasmaster_site/data/reaper_karne.json (2026-10-19 09:38) & data/kivilcim_karnesi.jsonl (2026-10-19 08:37) |
K062 · Training Set Candidate: KÜTÜK-5 SHA-256 Mirror Twin, 98.6% Bureaucracy Trap & N=96 Synaptic Harvest[training-set-candidate] | Empirical record of SHA-256 duplicate log detection, 98.6% bureaucracy trap pruning, and N=96 weight harvest | due: 2026-11-03 | blog/kim-kimi-egitiyor.html:351 data/hasat_defteri.jsonl (2026-10-20 23:41) & data/denetci_2x2.jsonl |
K063 · Training Set Candidate: Digital Dentistry 20 Field Verdicts & 10 External-Voice Sector Synthesis[training-set-candidate] | 20 clinical field verdicts and 10-voice sector synthesis backed by 360/394 Reaper-verified citations | due: 2026-11-02 | blog/dijital-dis-hekimligi-olan-ve-olmasi-gereken.html:417 atlasmaster_site/data/reaper_karne.json (2026-10-19 09:38) |
K074 · Training Set Candidate: 20-Question Evidence-Backed Debate with Maya (Counter-Thesis + Falsifier Set)[training-set-candidate] | 20 structured clinical and architectural questions with explicit counter-theses and falsification criteria | due: 2026-11-02 | bot/maya/index.html:415 atlasmaster_site/bot/maya/tartisma-20.json (2026-10-19 04:00) |
2. Stamp Staleness Gate (14-Day Rule & Archive Contract)
- Automated Generation: This page is never edited by hand; whenever an item's stamp or measurement date changes in
envanter_1021.json,uret_fayda_defteri.pyrebuilds the ledger. - Hourly/Daily Guard (
site_bakim.py·damga_bayat): Every measured item carries ason_olcum + 14 daysvalidity window. Expired items trigger adamga: K0xx vadesi geçtialert. - Zero Deletion, Transparent Archive: The
14pending and12unverified items that do not gain primary evidence within14days (by2026-11-04) are moved to the archive layer without deleting their inventory IDs. - No Clinical Claim Boundary: All dental and 3D cephalometric entries in this ledger represent computational design and optical/geometric simulation measurements only; clinical diagnosis and clinical decisions belong solely to the dentist.
3. Frequently Asked Questions (Benefit Ledger & 14-Day Rule)
- Which items are included in the Benefit Ledger table and which are excluded?
- Out of
75total inventory items, only the49measured items backed by a primary file, log, or self-test script appear in the table. The14items awaiting measurement and12unverified items are reported strictly as aggregate counts and excluded by name. - How does the 14-day validity and archive gate work?
- Each item has a
14-day validity deadline (2026-11-04) from its last measurement timestamp. Thesite_bakim.pyguard checks all deadlines via thedamga_bayatgate and archives unmeasured items without deletion. - How are the primary evidence links verified?
- Every row links directly to the corresponding
atlasmaster.apppage,bot/olgular.jsonentry, orreaper-en.html#karnescorecard, andkanit_fayda/selftest_fayda_1021.pymechanically verifies that the displayed item IDs equal the measured set.