Monogate — Evidence & Claims Status

Held lanes

private-atlas-v0 — held by atlas_a51_private_atlas_reviewer_response_hold_selector_feed — held_pending_reviewer_response_or_explicit_redirect

Resolved since the last snapshot (owner review, 2026-07-05): public-math-review, training-cost-estimator. One private lane was also removed — it should never have surfaced publicly.

electronics-inbox: the laptop agent's 8 live-hardware captures (ARTY-001–007, real Arty A7-100T silicon) were filed and validated by ee_bridge_a4 — 7/7 accepted, ready_for_private_review. Canonical aggregation (EH-A7 → PUB-R0 → PUB-R1) regeneration to follow.

Retracted claims

17 claims retracted and kept as faithful history — the ledger never silently drops them.

Show the 17 retracted claims

claim:C259_cross_genome_3 — exploration/C259_cross_genome/FINDINGS.md:93 — ## Falsifiability — what would have killed the claim
claim:CapCard_Distributed_Evidence_Quarantine_2026_05_16_2 — exploration/CapCard_Distributed_Evidence_Quarantine_2026_05_16/FINDINGS.md:13 — - A summary file that records source rows, accepted rows, rejected rows, and boundary counts.
claim:CapCard_Distributed_Evidence_Review_Gate_2026_05_16_1 — exploration/CapCard_Distributed_Evidence_Review_Gate_2026_05_16/FINDINGS.md:25 — - rows that should be rejected
claim:CapCard_Distributed_Evidence_Revision_Review_2026_05_16_1 — exploration/CapCard_Distributed_Evidence_Revision_Review_2026_05_16/FINDINGS.md:21 — - Revised rows rejected: 0
claim:CapCard_Distributed_Lifecycle_Closeout_2026_05_16_1 — exploration/CapCard_Distributed_Lifecycle_Closeout_2026_05_16/FINDINGS.md:14 — - rejected rows: 0
claim:CapCard_Distributed_Lifecycle_Control_Panel_2026_05_16_1 — exploration/CapCard_Distributed_Lifecycle_Control_Panel_2026_05_16/FINDINGS.md:12 — - Counts accepted, revision-needed, and rejected rows.
claim:CapCard_Distributed_Reviewed_Evidence_Ledger_2026_05_16_1 — exploration/CapCard_Distributed_Reviewed_Evidence_Ledger_2026_05_16/FINDINGS.md:14 — - Rejected rows JSONL.
claim:CapCard_Distributed_Reviewed_Evidence_Ledger_2026_05_16_2 — exploration/CapCard_Distributed_Reviewed_Evidence_Ledger_2026_05_16/FINDINGS.md:23 — Rejected rows remain useful as negative evidence.
claim:E196_algorithmic_corpus_5 — exploration/E196_algorithmic_corpus/FINDINGS.md:116 — prediction in the session prompt is rejected. Real-world
claim:E196_algorithmic_corpus_16 — exploration/E196_algorithmic_corpus/FINDINGS.md:224 — **Verdict: REJECTED.** Median `max_path_r` is 1.0 for BOTH groups.
claim:EML_Puzzle_Curriculum_Evidence_Ledger_2026_05_16_1 — exploration/EML_Puzzle_Curriculum_Evidence_Ledger_2026_05_16/FINDINGS.md:8 — - Rejected: 0
claim:EML_Puzzle_Evidence_Ledger_2026_05_16_1 — exploration/EML_Puzzle_Evidence_Ledger_2026_05_16/FINDINGS.md:11 — - Rejected: 0
claim:Frontier_A_resolution_096_097_2026_05_11_1 — exploration/Frontier_A_resolution_096_097_2026_05_11/FINDINGS.md:70 — guess to 2×10⁻⁵ /t is **RETRACTED.**
claim:Frontier_A_sigma099_T1e6_2026_05_10_1 — exploration/Frontier_A_sigma099_T1e6_2026_05_10/FINDINGS.md:10 — (PARTIAL_OBSERVATIONS only — killed to free CPU for this run).
claim:Frontier_A_sigma099_T1e6_2026_05_10_4 — exploration/Frontier_A_sigma099_T1e6_2026_05_10/FINDINGS.md:43 — killed T=10⁵ probe earlier today.
claim:Frontier_D_hd_vs_chain_2026_05_10_1 — exploration/Frontier_D_hd_vs_chain_2026_05_10/FINDINGS.md:99 — spectrum) is **rejected at this resolution and corpus size**. The
claim:PETAL_Local_Attempt_Ingestion_Simulator_2026_05_16_1 — exploration/PETAL_Local_Attempt_Ingestion_Simulator_2026_05_16/FINDINGS.md:10 — - Invalid local cases rejected: 3
claim:Qwen_Puzzle_Curriculum_Pack_Evidence_Ledger_2026_05_16_1 — exploration/Qwen_Puzzle_Curriculum_Pack_Evidence_Ledger_2026_05_16/FINDINGS.md:8 — - Rejected: 0
claim:alpha-6.214-recheck-2026-04-27_2 — exploration/alpha-6.214-recheck-2026-04-27/FINDINGS.md:5 — **Status:** REJECTED-IN-CURRENT-SETUP (the value does not appear at any

Negative results

claim:E193_numerical_stability_1 — exploration/E193_numerical_stability/FINDINGS.md:5 — **Outcome.** Partial signal (good case). Tool prototype NOT shipped.
claim:E193_numerical_stability_3 — exploration/E193_numerical_stability/FINDINGS.md:108 — negative, **two-sided p = 0.69 (not significant)**.
claim:E193_numerical_stability_4 — exploration/E193_numerical_stability/FINDINGS.md:162 — (NOT significant at alpha=0.05). Sample size at c=-1 is small (n=19).
claim:E199_superbest_stability_2 — exploration/E199_superbest_stability/FINDINGS.md:11 — tested. **30% accuracy is below the 70% threshold for a generic
claim:cat_vision_2 — exploration/cat_vision/FINDINGS.md:56 — intensity. Below human cone threshold; only rod-driven luminance
claim:forge_lean_gate_accepted_sorry — forge/tools/equivalence/lean_runner.py — Our own Lean equivalence gate treated compiled as proven — but a sorry compiles with returncode 0 (Lean only warns). A full-corpus #print axioms scan found two kernels, smoothstep and attitude_step, that had been passing as verified with an undischarged obligation. Gate fixed to reject sorry; both are now honestly reported as deferred (6cd73f6). The certificate's (sorryAx-free) line, previously a hardcoded string, now runs a real #print axioms check (ac962a1).

Standing claim rule

No training-cost estimate, training-savings, estimator-accuracy, runtime performance, compiler-correctness, SDK-stability, hardware-readiness, silicon-readiness, catalog-completeness, or broad EML-advantage claim unless a bounded artifact proves that exact claim.

MachLib verification status (CI-generated)

MachLib verification status

Fetching latest status.json from agent-maestro/machlib@status-data…

JavaScript is required to render the CI-emitted verification status. The raw data is at raw.githubusercontent.com/agent-maestro/machlib/status-data/status.json and may be inspected directly.

The block above is fetched fresh from the machlib repo's CI-only status-data branch on every page load. The displayed SHA + timestamp belong to the data, not this page. Hand-edits to the source tree cannot reach this surface; only the GitHub Actions runner writes the branch.

Other Lean surfaces (still byte-derived)

Other Lean surfaces

Lean theorems: 470 (monogate-lean MonogateEML/, byte-derived 2026-06-10)
Lean sorries: 5 (aggregator-verified, monogate-lean, byte-derived 2026-06-10)

These two are still byte-derived from pub-r0-brake-side-ledger-generator (2026-06-10) rather than CI-emitted. Folding them into the CI-generated status.json is follow-up work; until then, they may lag.