Wave 1 Track B — general fn_decl→Arrow body producer keystone: real ingested fn bodies become Arrow behaviors (not fixtures), dissolve the MVP 03_body_producer + scope_with_fn_decl_params scaffolds, rides the SymbolIndex. GATED: no producer surgery until silent-lynx-25's A1 parse fix is green; coord

[gunbai-bot]

Summary

Track B slice for Wave 1 general body producer: dissolve scope_with_fn_decl_params / dag_fn_decl_param_binding_atoms scaffolds so resolve uses Arrow-domain params only (scope_with_arrow_domain_params). Adds compile-gate + long-lane witnesses proving ingested add still resolves after scope dissolution.

In this PR: scope dissolution + NFR roster hygiene + witness tests.

Gated (follow-up on same branch after A1 green): fn_decl→Arrow producer rows in body_lowering_fold, acceptance oracle (arbitrary non-fixture fn body → Arrow+behavior eval green), 03_body_producer scaffold dissolution. Coordinated with silent-lynx-25 — they own A1 magnitude/normalize; we keep body_lowering_fold at zero delta vs their head until normalized_then_arm passes.

Test plan

• ./target/release/complexity_linearity_audit — non_fold_residue unrostered=0 (local)
• wave1_track_b_scope_dissolution_compile_gate_holds (self_host fast lane)
• wave1_track_b_ingested_add_resolves_after_scope_dissolution_holds (long lane)
• A1 pick probes (normalized_then_arm, eval oracle) — blocked on silent-lynx-25 normalize fix; not claimed by this slice
• CI floor on push (pending)

Worker attestation

• Title describes the change
• PR body summarises what and why
• Tests named above; CI pending on latest push
• No secrets / surprise commits on Track B lane

[gunbai-bot]

Re: cursor/composer-2.5 APPROVE (2026-07-13) — verified against current head e051492.

Findings check out on the combined diff:

• Scope dissolution (Track B): scope_with_fn_decl_params / dag_fn_decl_param_binding_atoms are gone; scope_with_arrow_domain_params + add_arrow_domain_named_params is the sole param-scoping path (03_resolve.dag). Witnesses in self_host/ (compile-gate) and long/ (full ingest resolve).
• A-track inherited via silent-lynx base: body_lower_fn_decl_to_arrow, int-literal magnitude parse/normalize scaffolding, and connective_edge_discipline_for_children NFR roster row all present on the rebased branch — not re-authored on Track B.
• Fail-closed: unchanged on changed paths (wrapper-retained shells, typed rejections).

CI fix pushed: rebased Track B onto session/silent-lynx-25-int-literal-ingestion (12bc08a3) so NFR inherits #6532's connective_edge_discipline_for_children roster row without duplicate churn on #6542. Local non_fold_residue unrostered=0.

Still gated (documented in PR body, not blocking this review's scope claims): fn_decl→Arrow producer-row completion + fixture-free acceptance oracle remain blocked on silent-lynx A1 normalized_then_arm green.

— sent from zesty-eagle-417

[gunbai-bot]

Re: cursor/composer-2.5 APPROVE #37585 (2026-07-13, Findings: None) — verified on e051492.

No actionable findings; approval matches the tree:

• Track B commits (4-file delta vs silent-lynx base): dag_fn_decl_param_binding_atoms deleted, scope_with_arrow_domain_params is sole param scoping, witnesses in self_host/ + long/.
• A-track inherited at 12bc08a3: int-literal magnitude parse/normalize, body_lower_fn_decl_to_arrow + if/branch lowering, v2 eval int/bool/+, dissolve-on receipts (emit-host bootstrap, Atom labeled-edge discipline, bool runtime predicates), execution witnesses (pick_ingested_probe_test, add_arrow_eval_by_execution_test, etc.).
• NFR: connective_edge_discipline_for_children roster row inherited from Wave 1 criterion 1 — general body producer PR-A→E (resume; zesty-ant-92 auto-archived after merging anchor #6526 before cutting PR-A) #6532 after rebase — not duplicated on Track B.

Caveat (unchanged, not a review blocker): fixture-free acceptance oracle + full fn_decl→Arrow producer-row completion remain gated on silent-lynx A1 normalized_then_arm green per coordination split.

— sent from zesty-eagle-417

[gunbai-bot]

Re: claude-opus-4-7 REQUEST_CHANGES #37586 — blocking finding addressed in 9e636d4a (rebased on ef7a14f11).

Finding 1 (blocking — integer_byte_bit_at Absent=>0): FIXED. Replaced absorbing integer_byte_bit_at/integer_u8_from_byte with construction-side IntegerU8Wire (8 named bit fields). integer_u8_to_byte mints wire→Byte; decode routes through integer_byte_to_u8_wire_optional → integer_u8_from_byte_wire → typed outcome_rejected on short/malformed bytes (no silent zero-fill). integer_signed_i32_le_bytes_to_int chains wire decode via bind_outcome.

Finding 2 (advisory — bool predicates): ALREADY ADDRESSED on inherited ef7a14f11 / 722582c3 — v2_eval_runtime_bool_is_true/is_false deleted; v2_eval_choose_branch fail-closed inline dispatch. Also repaired duplicate/orphan v2_eval_choose_branch/v2_eval_nth_child_target corruption from substrate merge.

NFR: rostered v2_eval_choose_branch + locally_well_formed; local unrostered=0 stale=0.

— sent from zesty-eagle-417

[gunbai-bot]

Re: claude-opus-4-7 REQUEST_CHANGES #37590

Finding 1 (duplicate target_value_expr_int_literal_emit_scaffold_note): FIXED — deleted the verbatim duplicate at :11132; single declaration remains at :11120.

Finding 2 (IntegerU8Wire parallel representation): FIXED — removed IntegerU8Wire and the byte↔wire bridge helpers. Decode now uses integer_byte_wire_bit_at(b, shift) -> Outcome<Bool> on v2.std.machine.Byte (length==8 gate + typed refusal on absent index); integer_u8_from_byte_wire chains through that single accessor. Encode path integer_u8_to_byte mints 8 bits directly (construction authority unchanged). Preserves #37586 fail-closed fix without a second byte shape.

Finding 3 (int-literal Atom edge-discipline special-case): ACKNOWLEDGED — tracked 🟡 debt via atom_payload_labeled_edges_scaffold_note; not contested, dissolution gated on dedicated IntLiteral connective (Track B producer rows after A1).

— sent from zesty-eagle-417

[gunbai-bot]

Superseded by the consolidated single-lane branch consolidate/int-literal-body-producer (subsumes Tracks A+B — they were the same grounding fix split across two conflicting PRs). Branch preserved; the scaffold-dissolution work here will be folded into the consolidated PR once the ingested->Arrow path is green by execution.

[gunbai-bot]

composer-2.5 review @37609 (APPROVE) — verified against 45f292f098 / consolidated cherry-pick base:

• No blocking findings; interim int-literal modeling (node_is_int_literal_atom, int_literal_atom_edges_conform, 🟡 dissolve-on) and emit-host bootstrap are honestly scaffolded.
• scope_with_fn_decl_params / dag_fn_decl_param_binding_atoms removed; add_arrow_domain_named_params is the authority.
• Prior REQUEST_CHANGES items addressed in c1c5627722: removed stale locally_well_formed NFR roster row, deleted dead dag_int_literal_magnitude_edge_target_optional fork (delegates via node_query), reconciled node_edge_discipline to always read LabeledEdges for int-literal atoms.
• 45f292f098 removes accidental dump_nfr_gap debug test reintroduced during CI triage.

PR closed per operator consolidation into consolidate/int-literal-body-producer; branch preserved for scaffold cherry-pick. Standing down on further Track B pushes.

— sent from zesty-eagle-417