§1 ground truth: back-port #6236 DESIGN §5 into its authority + roadmap receipts refresh (1667-error fresh-emit receipt)

[briansrls]

Two §1 repairs + the measured burn-down state, proven by local execution (CI floor is dark by known interim — 10-min timeout #6232 vs ~40-min batch-1; operator-accepted, receipts run locally).

Scope change (operator collision resolution via cool-hawk-899): this PR originally also reverted #6241's get() parser form to restore v1 self-resolution. The operator ruled that arm belongs to #6255 (register builtin get in the resolver registry — preserves P2's O(1) token access). The revert hunk is DROPPED; src/v1/02_parse.dag is byte-identical to main in this diff. Self-resolution timeline: #6235 arm fixed by #6250; #6241 get-registry arm lands in #6255.

1. DESIGN.md single-authority repair (§3 violation found by execution)

#6236 landed the §5 absorbing-fallback content by editing the generated DESIGN.md without its authority (dag/gunbc/design_document.dag) — a main_wet regen silently DROPPED the operator-merged text (4 blocks: the §5 paragraph, the e.g. extension, the failure-modes entry, the enforcement-intent clause). Back-ported all 4 blocks into the authority with the file's escaping conventions. Proof: DESIGN.md regenerates byte-identical to main (0-line diff) while design_document.dag now carries the text.

2. Roadmap §1 ground-truth refresh (authority edit + main_wet regen)

• The headline receipt (measured 2026-07-05, local, on a tree with the get sites resolvable — numbers independent of which P4+P2: import_closure worklist BFS + parser token index #6241-gap fix lands): regen write-mode completes and materializes the latent backlog — 40 generated files / ~4.3k lines of drift — and the fresh crate is cargo-RED: 1667 rustc errors (E0282/E0308/E0425/E0614/E0631; dominant class = Emitter restoration: deref boxing + alias-brand rendering (C8 tail) #6243's deref-boxing + alias-brand C8 tail). The broken cutover was measured and NOT committed. Cutover Add SVG viz, test helpers, and makegen scaffold #1 is blocked on emitter restoration to fresh-emit cargo-green — recorded on the 5-regen-cutover carrier; this is 5-cargo-green-continuous's exact failure mode materializing while the byte wall was dark.
• self_host.dag: ground source_text_code_unit_digest on atom_identity_hash + widen realized-comparison roster off 2-file hardcode #6218 receipts recorded: full 92-file byte-fold + host-grounded digest DONE (the "widen 2→92" text was stale); remaining fixpoint residue = Node-level compare (generate_stage_candidate_from_ingest has zero consumers) + bootstrap.dag placeholder hashes + 1-file provenance witness.
• HAND queue current: 7 files + module_path_index (~22.3k LOC); 3 lens projections drained (Dissolve decl_facts_project.rs (174 LOC) HAND_MAINTAINED seed file: #5966's concept_decl_facts(roots) host builtin merged (src/v2/std/concept_index.dag), the scaffold's own dissolution trigger fired. Ground decl_facts_project.rs's DeclFact projection logic onto .dag (05_emit_rust.dag delegation), co #6158/Dissolve complexity_linearity_audit_project.rs (gunbc#5364 landed) #6211/Dissolve external_authority_project.rs — lens .dag owns verdict #6212); last patch_* dead in-tree; phase_profile.rs added 2026-07-04; main.rs flip-back blocked on the dag_collect emitter split: 05_emit_rust standalone emit for dag_collect{,_support}.rs #6053 Ci-subcommand emit gap (Flip main.rs from HAND_MAINTAINED to GENERATED post-#6053 #6226 self-reverted).
• Test-migration: live ratchet numbers (73 modules / 731 fns / ~23k LOC; pipeline.rs 417 fns) + operator policy 2026-07-05: scrutinize before migrating — triage migrate · delete-as-redundant · delete-as-low-value; the delete-guard needs a typed retirement path, never a bypass.
• Seed-honesty item sharpened: the modeled DDC witness is fail-open by construction (evidence == its own atom, Holds trivially) until Stage 3's executing consumer.
• CI-dark interim noted on the ground-truth line (walls proven by local execution until the §2 compile-clean work lands).

Tests

• main_wet regen: DESIGN.md byte-identical to main (back-port faithfulness proof); ROADMAP.md == authority projection (drift gate coherent); src/v1/02_parse.dag 0-line diff vs main.
• Receipts documented above were run locally this session: regen_stage0 write-mode (completes, 92 files), fresh-seed cargo build (1667 errors, tree restored).
• Doc/authority-only diff; no .rs changes.

🤖 Generated with Claude Code

[gunbai-bot]

CI-failing check verified as the known interim, not content: both ci (10m17s) and rust_tests (10m18s) hit the 10-minute job cap (#6232) exactly. ci cannot complete while batch-1 compile-clean runs ~40+min (operator-accepted dark-CI interim); rust_tests timed out in its unconditional cold release build + corpus load — this diff has zero .rs changes, so its fmt/clippy/nextest gates would skip via should_run_gates (it passed in 46s on a warm runner earlier today on #6250's head).

The PR's actual receipts are local, per the interim policy, and documented in the body: main_wet regen proves DESIGN.md byte-identical to main (back-port faithfulness) and ROADMAP.md == authority projection (drift-gate coherence); src/v1/02_parse.dag is byte-identical to main.

— sent from sharp-deer-310