Shell→dag P2b consumer slice (successor to archived bold-pike-629; plan agreed on-record): (1) route converge_gunbc_pinned_tree fresh_standup fragment through emit(intent,Bash) via procedure/cmdsubst_assign/let_assign ONLY (steady-state functions untouched — P4 lane owns those); (2) byte-identity wi by briansrls · Pull Request #6573 · gunb-ai/gunbc

briansrls · 2026-07-14T03:41:18Z

Auto-opened by session-dashboard for session neat-boar-644.
Pushing to session/neat-boar-644 advances this PR.

Worker attestation

Before flipping this PR to ready for review, confirm each item:

Title describes the change (not the session id or branch).
PR body summarises what and why (replace the TODO below).
Tests run: name the command (e.g. npm test, cargo test) and the result.
If this closes a work item, the body contains a Closes #N directive.
No commits on this branch are surprises (no fork/cherry-pick I did not make).
No secrets / credentials / large binaries staged.

Summary

TODO: replace this paragraph with one or two sentences naming the change and its motivation. Reviewers read this first.

Test plan

TODO: list the commands that ran (or "no tests changed; relied on CI") and the outcome.

…9; plan

gunbai-bot · 2026-07-14T04:34:52Z

Review responses (neat-boar-644)

claude/claude-opus-4-7 (APPROVE) and cursor/composer-2.5 (APPROVE) — verified against current HEAD 501ff5ed. No code changes required from either review; findings are accurate.

Follow-up commit 501ff5ed completes the parent brief gates that landed after those reviews:

Gate 2: fresh_standup_bootstrap_arm_golden() byte-identity + witness_fresh_standup_bootstrap_arm_perturb_red
Gate 3: deleted fleet_converge_fresh_standup_bootstrap_emit_frontier census rows
Gate 4: removed orch_construct_arith / realize_arith / arith grammar + witnesses (no pre-runtime live site)

fleet_converge_emit_holds and tier-2 fresh-standup witnesses pass locally via claim_batch.

— sent from neat-boar-644

gunbai-bot · 2026-07-14T04:35:43Z

cursor/composer-2.5 review @ 501ff5ed (37732) — verified against current HEAD; no code changes required.

Spot-checks match the review:

fresh_standup_bootstrap_emit_step poison path at fleet_converge_emit.dag:63-71 with absence asserted in witness_fresh_standup_bootstrap_emit_bytes_hold
fleet_converge_fresh_standup_bootstrap_emit_dissolution_trigger scaffold replacing the deleted frontier census
witness_fresh_standup_bootstrap_arm_perturb_red perturb RED at fleet_converge_emit_test.dag:329-331
Steady-state $((…)) sites (e.g. line ~221) remain hand literals for P4
ExprArith / orch_construct_arith fully removed; no corpus references remain

CI floor job still running on this SHA; merge blocked only on checks pending.

— sent from neat-boar-644

gunbai-bot · 2026-07-14T04:35:53Z

claude/claude-opus-4-7 review @ 501ff5ed (37737) — verified against current HEAD; no code changes required.

Confirmed:

Four fresh_standup fragments routed through orch_emit_step / bash_orchestration_emit_medium() with poison-on-refusal (fleet_converge_emit.dag:63-71)
Byte-identity + perturb RED witnesses (witness_fresh_standup_bootstrap_emit_bytes_hold, witness_fresh_standup_bootstrap_arm_perturb_red)
Frontier census deleted; fleet_converge_fresh_standup_bootstrap_emit_dissolution_trigger scaffold in place
ExprArith / orch_construct_arith / ArithAdd|Sub fully removed — zero corpus references remain (git grep clean on HEAD)

CI floor job still pending on this SHA.

— sent from neat-boar-644

…_emit_dissolution_trigger. The prior bind targeted a non-existent frontier list; re-point to the #6573 dissolution row and tighten witness to assert module_path + decl_name resolution. Co-authored-by: Cursor <cursoragent@cursor.com>

…rge_apply (#6572) * WIP: Shell→dag P4 KEYSTONE: mint HostEffect::ConvergePlan{policy: HostConverg * WIP: Shell→dag P4 KEYSTONE: mint HostEffect::ConvergePlan{policy: HostConverg * Shell→dag P4 KEYSTONE: mint ConvergePlan effect, EmitArtifactThenThinRun transport, and converge_apply consumer. Adds typed HostEffect::ConvergePlan with ConvergeOnHost cell resolution, fail-closed per-knob ConvergeKnob interpreter frontier, fleet_converge_apply converge_apply() entry point, and witness coverage for the new apply path. Co-authored-by: Cursor <cursoragent@cursor.com> * Fix CI: exhaust ConvergePlanApplied in floor_diff_observe evidence matches. The new HostEffectEvidence arm must be handled fail-closed in git-diff observation reconciliations so compile-clean gate resolves. Co-authored-by: Cursor <cursoragent@cursor.com> * WIP: Shell→dag P4 KEYSTONE: mint HostEffect::ConvergePlan{policy: HostConverg * WIP: Shell→dag P4 KEYSTONE: mint HostEffect::ConvergePlan{policy: HostConverg * Address review #37742: scaffold medium-as-string carriers, single ConvergePlan authority. Add Disposition scaffold rows for BootstrapFragment, ThinInvocation, and EmitArtifactThenThinRun bound to bash-emit dissolution targets; drop redundant ConvergePlan.host so policy.identity is the sole authority (target checked at resolve only). Co-authored-by: Cursor <cursoragent@cursor.com> * Fix bootstrap scaffold bind to fleet_converge_fresh_standup_bootstrap_emit_dissolution_trigger. The prior bind targeted a non-existent frontier list; re-point to the #6573 dissolution row and tighten witness to assert module_path + decl_name resolution. Co-authored-by: Cursor <cursoragent@cursor.com> * Deduplicate fleet host lookup in converge_apply_for_host. Single fleet_compute_host_for_identity per host; converge_plan_effect is now a pure policy→effect mapper with no redundant scan. Co-authored-by: Cursor <cursoragent@cursor.com> --------- Co-authored-by: Brian Searls <briansearls1@gmail.com> Co-authored-by: Cursor <cursoragent@cursor.com>

The nightly affected-set-falsifier's whole-corpus cold sweep (gunbc_falsifier_batches, SelectionPredictOnly) surfaced 4 red live-tree witnesses = 2 real drift debts of the documented masking class (PRs landed changes the per-PR affected-set selection did not run these live-tree witnesses for, so the red surfaced on the next cold sweep, not the landing PR — 4th instance, siblings to #6530/#6533): 1. doc_graph_has_no_orphan_docs (x2): docs/plans/live-read-witness-classification-design.md (added #6560/#6582) was referenced only in prose data:String notes, never a doc-graph edge. Fix: add a `bind:` provenance row in the owning lens file (the established convention, cf. commit_workflow_plan_doc_provenance) -> doc-graph root, orphans 1->0. 2. non_fold_residue_clean_holds (x2): resolved report = 3 unrostered, 0 stale. Backfill NON_FOLD_RESIDUE_ROSTER with the 3 sites (dated comment + per-site dissolution triggers): orch_emit_let_step (#6573; dissolves with the backward grammar-row emit fold), live_read_carrier_eq / path_pattern_eq (#6582; structural-equality nests, dissolve with derived == from inhabitance) — all matching existing roster precedent. Verified green-by-execution: all 4 previously-red witnesses (+6 siblings) PASS via claim_batch on a fresh v1-compiler build. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* WIP: affected sest falsifier * Affected-set falsifier burn-down: link orphan doc + backfill 3 nfr sites The nightly affected-set-falsifier's whole-corpus cold sweep (gunbc_falsifier_batches, SelectionPredictOnly) surfaced 4 red live-tree witnesses = 2 real drift debts of the documented masking class (PRs landed changes the per-PR affected-set selection did not run these live-tree witnesses for, so the red surfaced on the next cold sweep, not the landing PR — 4th instance, siblings to #6530/#6533): 1. doc_graph_has_no_orphan_docs (x2): docs/plans/live-read-witness-classification-design.md (added #6560/#6582) was referenced only in prose data:String notes, never a doc-graph edge. Fix: add a `bind:` provenance row in the owning lens file (the established convention, cf. commit_workflow_plan_doc_provenance) -> doc-graph root, orphans 1->0. 2. non_fold_residue_clean_holds (x2): resolved report = 3 unrostered, 0 stale. Backfill NON_FOLD_RESIDUE_ROSTER with the 3 sites (dated comment + per-site dissolution triggers): orch_emit_let_step (#6573; dissolves with the backward grammar-row emit fold), live_read_carrier_eq / path_pattern_eq (#6582; structural-equality nests, dissolve with derived == from inhabitance) — all matching existing roster precedent. Verified green-by-execution: all 4 previously-red witnesses (+6 siblings) PASS via claim_batch on a fresh v1-compiler build. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Brian Searls <briansearls1@gmail.com> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…spatch onto finished work) Wave C4's brief described fleet_converge_emit.dag as ~275 lines of Doc-literal bash (12+ fn defs, for-loops, while-read drain, arithmetic, verdict if/elif) to route through EmitArtifactThenThinRun. That state no longer exists: the keystone landed as #6572 (ConvergePlan + EmitArtifactThenThinRun + converge_apply), #6585 (committed .github/fleet-converge.sh golden shrunk 323 -> 21 lines), and #6598 (gunbc converge --host CLI receiver); FLAGs 2a(i)/2b/2c were signed/discharged 2026-07-14. Receipts re-verified by execution 2026-07-20: fleet_converge_emit_{bootstrap,thin_run,policy}_holds, fleet_converge_emit_holds, and fleet_converge_apply_holds all PASS, including the drift gate vs the committed golden and its one-byte perturb RED. The stale "~275 lines" framing has now produced two worker dispatches onto finished work (slice 0 on 2026-07-16, slice 2 on 2026-07-20 — this session). Mirroring #6878 (the Phase-0/1 landing-record precedent), this change updates the authority carriers so it cannot recur: - roadmap_authority.dag 6-shell-slice2: reworded to MERGED / operator-sign-off-PENDING with the landing PRs and receipt date (done stays false — the sign is the operator's; mirrors 6-shell-intent-phase1's shape). - plans/shell_emission_model.dag: section-1 status and the slice-2 row updated to the merged/receipt state; dispatch note added. - shell-to-dag-residual-census-and-arc-completion.md (hand-authored, no carrier): category-C fleet_converge_emit row marked dissolved with receipts; section-2 goal marked achieved (sign-off pending); sequence steps 1-2 marked landed, step 3 annotated still-open (the doc projection trigger remains a Scaffold binding fleet_converge_thin_invocation); P2 consumer slice marked landed (#6573); critical-path summary flipped from "P4 execution" to P5/P6 + the operator sign. - ROADMAP.md / docs/plans/shell-emission-model.md / .github/live-deploy-srv1-apply.sh regenerated via generated_artifact_gate main_wet; drift gate green (run_generated_artifact_drift_gate ExitSuccess), and roadmap_authority_witnesses + roadmap_page_keystone_holds PASS. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

…closeout: keystone already merged as #6572/#6585/#6598) (#6948) * WIP: Wave C4 (shell->dag): fleet_converge_emit (~275 lines Doc-literal bash: * Record shell-emission slice-2 landing in authority carriers (no re-dispatch onto finished work) Wave C4's brief described fleet_converge_emit.dag as ~275 lines of Doc-literal bash (12+ fn defs, for-loops, while-read drain, arithmetic, verdict if/elif) to route through EmitArtifactThenThinRun. That state no longer exists: the keystone landed as #6572 (ConvergePlan + EmitArtifactThenThinRun + converge_apply), #6585 (committed .github/fleet-converge.sh golden shrunk 323 -> 21 lines), and #6598 (gunbc converge --host CLI receiver); FLAGs 2a(i)/2b/2c were signed/discharged 2026-07-14. Receipts re-verified by execution 2026-07-20: fleet_converge_emit_{bootstrap,thin_run,policy}_holds, fleet_converge_emit_holds, and fleet_converge_apply_holds all PASS, including the drift gate vs the committed golden and its one-byte perturb RED. The stale "~275 lines" framing has now produced two worker dispatches onto finished work (slice 0 on 2026-07-16, slice 2 on 2026-07-20 — this session). Mirroring #6878 (the Phase-0/1 landing-record precedent), this change updates the authority carriers so it cannot recur: - roadmap_authority.dag 6-shell-slice2: reworded to MERGED / operator-sign-off-PENDING with the landing PRs and receipt date (done stays false — the sign is the operator's; mirrors 6-shell-intent-phase1's shape). - plans/shell_emission_model.dag: section-1 status and the slice-2 row updated to the merged/receipt state; dispatch note added. - shell-to-dag-residual-census-and-arc-completion.md (hand-authored, no carrier): category-C fleet_converge_emit row marked dissolved with receipts; section-2 goal marked achieved (sign-off pending); sequence steps 1-2 marked landed, step 3 annotated still-open (the doc projection trigger remains a Scaffold binding fleet_converge_thin_invocation); P2 consumer slice marked landed (#6573); critical-path summary flipped from "P4 execution" to P5/P6 + the operator sign. - ROADMAP.md / docs/plans/shell-emission-model.md / .github/live-deploy-srv1-apply.sh regenerated via generated_artifact_gate main_wet; drift gate green (run_generated_artifact_drift_gate ExitSuccess), and roadmap_authority_witnesses + roadmap_page_keystone_holds PASS. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: Wave C4 (shell->dag): fleet_converge_emit (~275 lines Doc-literal bash: * Merge main into session/swift-moth-345; regenerate live-deploy artifact from merged carriers Same class as the prior merge: the only conflict was the generated .github/live-deploy-srv1-apply.sh (embeds ROADMAP.md; both sides regenerated it). Took main's copy, rebuilt gunbc from the merged seed (main changed cli_run.rs), re-ran generated_artifact_gate main_wet, and verified the drift gate green (ExitSuccess). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> --------- Co-authored-by: Brian Searls <briansrls@gunb.ai> Co-authored-by: Claude Fable 5 <noreply@anthropic.com>

Brian Searls added 2 commits July 14, 2026 03:41

WIP: Shell→dag P2b consumer slice (successor to archived bold-pike-62…

059f76a

…9; plan

WIP: Shell→dag P2b consumer slice (successor to archived bold-pike-62…

ba45123

…9; plan

gunbai-bot Bot marked this pull request as ready for review July 14, 2026 04:06

WIP: Shell→dag P2b consumer slice (successor to archived bold-pike-62…

501ff5e

…9; plan

briansrls merged commit b48b21f into main Jul 14, 2026
3 checks passed

briansrls deleted the session/neat-boar-644 branch July 14, 2026 04:55

gunbai-bot Bot pushed a commit that referenced this pull request Jul 14, 2026

Merge origin/main: absorb #6573 bootstrap routing and #6568 P1.

cb5ca61

gunbai-bot Bot mentioned this pull request Jul 20, 2026

Record shell-emission slice-2 landing in authority carriers (Wave C4 closeout: keystone already merged as #6572/#6585/#6598) #6948

Merged

gunbai-bot Bot mentioned this pull request Jul 21, 2026

fleet-converge: dissolve bootstrap emitter scaffold + byte-oracle witness #7007

Merged

3 tasks done

Navigation Menu

Repository navigation

briansrls commented Jul 14, 2026

Uh oh!

gunbai-bot Bot commented Jul 14, 2026

Uh oh!

gunbai-bot Bot commented Jul 14, 2026

Uh oh!

gunbai-bot Bot commented Jul 14, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

briansrls commented Jul 14, 2026

Worker attestation

Summary

Test plan

Uh oh!

gunbai-bot Bot commented Jul 14, 2026

Review responses (neat-boar-644)

Uh oh!

gunbai-bot Bot commented Jul 14, 2026

Uh oh!

gunbai-bot Bot commented Jul 14, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant