← NS-GSM / v0.7 · FELRA Formal Proof Bridge

NS-GSM · v0.7 FELRA Formal Proof Bridge 2026-08-28

NS_GSM FELRA Formal Proof Bridge v0.7

In this environment, FELRA was never actually connected. What v0.7 builds is the protocol and pipeline toward the external tool FELRA (confirmed version 1.8.1/main, Lean backend), but the sandbox this version was built in has no lake, lean, or felra executable (the report, verbatim: lake: unavailable / lean: unavailable / felra: unavailable). Only 5 bridge targets (D103.1–D103.5, the ‘first supported family’ subset of the 12 PROOF assets), producing 5 FIR→Lean translations — obligation.lean was inspected directly, and these are genuine Lean 4/Mathlib theorem statements, with the proof body replaced by an explicit placeholder NSGSM_PROOF_BODY_REQUIRED, not sorry. But the actual, ingested FELRA execution result: 0. FORMAL_PROOF[lean] receipts: 0. A genuine ‘fail-closed’ ingestion gate is defined (10 required conditions, including exact FIR/obligation identity matching, non-empty axiom auditing, and prohibitions on sorry/undecided/hash mismatch) for future use once a real Lean execution result is ingested — but no such execution took place this round.

Builds a translation pipeline toward FELRA/Lean and a fail-closed ingestion gate; 5 FIR→Lean translations are produced, and the actual Lean-verification result is 0. The report lists, in bold, what v0.7 does not claim, including: ‘that Lean kernel verification occurred in this sandbox; that any FORMAL_PROOF[lean] exists in the release evidence; that Coq support already exists; that the D103 local-algebra proofs prove the parent-problem’s NS regularity; that formalization is complete; that route completeness holds; that representation completeness holds; or that the formal-NS root is closed.’

Connections

Relationship to the other versions, stated as closely as possible in the document's own words, not my interpretation.

NS-GSM progressv0.7 (8 versions total: seed dataset + v0.1–v0.7)

Loading…