InterfaceSummary v0: std carrier + first consumer by gunbai-bot[bot] · Pull Request #6246 · gunb-ai/gunbc

gunbai-bot · 2026-07-04T23:49:25Z

Summary

InterfaceSummary v0: std carrier (dag/std/interface_summary.dag) with signature-only contracts (FLAG D: ContractAbsent | SignatureContract), per-declaration export fingerprints folded into a module rollup hash (FLAG B), all hashing on std.content_hash (no third hash surface). Derived projection lens (v2.lens.interface_summary) reads live declarations via new host builtin export_signature_facts (signature nodes only — no bodies). src/v2/std/interface_summary.dag is an imports-only re-export shim over std.interface_summary (zero second definitions). First consumer: firewall witness pair discriminating body-only vs signature edits.

Test plan

target/release/claim_batch --source-root dag --source-root src/v2 --wet --entry src/v2/test/claim/interface_summary/interface_summary_firewall_test.dag --function interface_summary_firewall_body_only_edit_is_green --function interface_summary_firewall_signature_edit_is_red — both PASS locally
cargo build -p v1-compiler --release --bin claim_executor — builds clean
PR stays draft until operator signs load-bearing host changes (below)

OPERATOR-SIGN-NEEDED

This PR adds host builtin export_signature_facts in load-bearing v1 seed files:

src/v1/04_method.dag — builtin registry row
src/v1/stage0/src/v1_interpreter.rs — interpreter dispatch
src/v1/stage0/src/coproduct_reflection.rs — marshal_fn_export_signature_node / eval_export_signature_facts (params + return Arrow, no body)
src/v1/stage0/src/v1_compiler_infer_method.rs — manual patch (generated; regen not run)

Why reflection was forced: InterfaceSummary v0 is signature-only (FLAG D). Existing decl_facts marshals fn items with body skeletons, so signature fingerprints would include implementation detail — wrong firewall semantics. With no resolver S2a integration in scope, the only path to live-tree export signature nodes is the same reflection seam as decl_facts / module_declaration_facts_live: a dedicated builtin that projects signature-shaped nodes per export. This follows the decl_facts precedent but is a new surface on the load-bearing interpreter/reflection layer — operator sign required before gh pr ready.

Coordination — `decl_facts(roots)` / #5966

export_signature_facts is adjacent to the decl_facts(roots) #5966 blocker named in the structural-quadratic-wall audit: same reflection machinery, narrower projection (signature nodes without bodies). This builtin may partially unblock that lane (roots-scoped signature facts without full body marshaling). Operator should review before this lands shape — especially if #5966 plans to generalize decl_facts rather than accumulate per-projection builtins.

By-execution receipts (verbatim)

Firewall witness reads the live tree via export_signature_facts / module_declaration_facts_live host intrinsics. Classified host-scaffold (floor:host_scaffold data note + _facts_live( classifier signal in interface_summary_firewall_test.dag) — never node-skipped on the CI floor.

GREEN — body-only edit, export fingerprint unchanged

$ target/release/gunbc run --source-root dag --source-root src/v2 \
    --entry src/v2/test/claim/interface_summary/interface_summary_firewall_test.dag \
    --function firewall_body_v1_fingerprint_receipt
running firewall_body_v1_fingerprint_receipt()...
7c54bb720f0264f7

$ target/release/gunbc run --source-root dag --source-root src/v2 \
    --entry src/v2/test/claim/interface_summary/interface_summary_firewall_test.dag \
    --function firewall_body_v2_fingerprint_receipt
running firewall_body_v2_fingerprint_receipt()...
7c54bb720f0264f7

$ target/release/claim_batch --source-root dag --source-root src/v2 --wet \
    --entry src/v2/test/claim/interface_summary/interface_summary_firewall_test.dag \
    --function interface_summary_firewall_body_only_edit_is_green
PASS interface_summary_firewall_body_only_edit_is_green
[witness] interface_summary_firewall_body_only_edit_is_green: 3ms subject=e2531cfbb5d1b10c eval_self=0.000ms

(firewall_body_v1.dag body { x + 1 } vs firewall_body_v2.dag body { x + 2 } — same signature fn firewall_gate(x: Int) -> Int, fingerprint 7c54bb720f0264f7 both sides.)

RED control — signature edit, export fingerprint changes

$ target/release/gunbc run --source-root dag --source-root src/v2 \
    --entry src/v2/test/claim/interface_summary/interface_summary_firewall_test.dag \
    --function firewall_sig_variant_fingerprint_receipt
running firewall_sig_variant_fingerprint_receipt()...
de2f79f14b466a7f

$ target/release/claim_batch --source-root dag --source-root src/v2 --wet \
    --entry src/v2/test/claim/interface_summary/interface_summary_firewall_test.dag \
    --function interface_summary_firewall_signature_edit_is_red
PASS interface_summary_firewall_signature_edit_is_red
[witness] interface_summary_firewall_signature_edit_is_red: 3ms subject=64bf446ef7fe18c4 eval_self=0.000ms

(firewall_sig_variant.dag changes return type to Bool — fingerprint de2f79f14b466a7f ≠ baseline 7c54bb720f0264f7.)

gunbai-bot · 2026-07-05T00:46:33Z

Verified each finding against current head (068b3d):

interface_summary_v0_dissolution_trigger — present at dag/std/interface_summary.dag:32; lens carries WallAfterGrounding { dissolves_to: SingleAuthority } at src/v2/lens/interface_summary.dag:219-221.
No unit-modeling triggers — carrier uses branded hash types + closed enums only; no bare-scalar domain-unit fields in added .dag.
Fail-closed Absent arms — body_only_edit_preserves_export_fingerprint / signature_edit_changes_export_fingerprint return false on any Absent fingerprint (lines 71-73, 89-90).
Firewall discrimination — fixtures: firewall_body_v1/_v2 same sig (Int→Int), different bodies; firewall_sig_variant flips return to Bool. Both witnesses wired; scoped claim_batch --wet PASS + fingerprint receipts in PR body (7c54bb720f0264f7 == 7c54bb720f0264f7 GREEN; de2f79f14b466a7f RED).
export_signature_facts registration — symmetric sibling of decl_facts in coproduct_reflection.rs; consistent rows in 04_method.dag, v1_interpreter.rs, v1_compiler_infer_method.rs, v2/std/decl_index.dag.

Minor observation (P5 receipt on hand-Rust): No separate per-file receipt beyond mirroring eval_decl_facts — agreed this is the one place strict P5 would ask. Already flagged in PR body OPERATOR-SIGN-NEEDED (load-bearing seed extension pending operator sign); draft held deliberately for that gate, not idle-automation clearance.

No code changes from this review — findings affirm current shape.

— sent from jolly-lark-525

gunbai-bot · 2026-07-05T04:07:36Z

Draft held deliberately — not abandoned.

Dashboard nudge acknowledged. This PR stays draft until:

OPERATOR-SIGN on the load-bearing host bridge (export_signature_facts in seed Rust — same class as decl_facts). Generic dashboard idle nudges are not operator sign; draft is the gate.
CI green on head — prior failure @ c04b1b4 was batch-2 generated_artifact_drift (ci.yml trailing newline) + floor_diff_observe resolve break from P1 emit fix: dag_collect O(n^2) fingerprint memo #6242 without Unbreak v1 self-resolution after #6242: hoist .dag comments to data rows + dag_collect registry dedup #6262. Fixed on 51a2aa3: merged main (Unbreak v1 self-resolution after #6242: hoist .dag comments to data rows + dag_collect registry dedup #6262 self-resolution + Un-red the floor drift gate: back-port #6236 into design_document.dag + regen ROADMAP.md #6264), regen DESIGN.md.

Will gh pr ready when checks pass and operator signs the host extension. Not closing — InterfaceSummary v0 carrier + firewall witness are live and scoped witnesses pass.

— sent from jolly-lark-525

gunbai-bot · 2026-07-05T04:18:36Z

Rebased onto main — merge conflict resolved (be7f152).

Squashed the 26-commit WIP history onto current main (7d9804, #6263 step-level CI budget). Dropped superseded dag_collect WIP + interim 60min job-timeout commits (already on main via #6242/#6262/#6263).

Two clean commits:

InterfaceSummary v0 carrier + lens + firewall witness + export_signature_facts host bridge
ci.yml regen from CI budget breach = step-level FAILURE, never cancellation silence (+ port #6236 into the DESIGN model) #6263 authority (committed workflow was stale vs ci_workflow.dag)

GitHub reports MERGEABLE. Still draft pending OPERATOR-SIGN on host bridge.

— sent from jolly-lark-525

Matches ci_workflow.dag authority (37e4ec): build step gets rust_gate 15min; claim_executor floor invoke gets 30min interim scaffold. Fixes Co-authored-by: Cursor <cursoragent@cursor.com> #6246 @ be7f152 step timeout during cold cargo build (run 28729269837).

Std carrier + lens + firewall witness + export_signature_facts host bridge. CI interim budgets (split build/floor/rust-gate, 30min build, floor_diff relative roots). Replayed cleanly — no unrelated main reverts. Co-authored-by: Cursor <cursoragent@cursor.com>

… receipt. - Add v2.std.decl_index::export_signature_facts_host_scaffold_dissolution_trigger (#5966 / decl_facts_for_roots census; dissolves with interface_summary_v0) - import_interface_hashes_for_module refuses on missing import hash (Absent); module_key_for_rel_path returns ModuleKey? - Witnesses: scaffold receipt cites #5966; unresolved import → Absent Addresses Composer REQUEST_CHANGES review 36013. Co-authored-by: Cursor <cursoragent@cursor.com>

gunbai-bot · 2026-07-06T10:47:40Z

Composer review 36013 (REQUEST_CHANGES) — addressed in latest push:

Hand-Rust receipt — added checkable export_signature_facts_host_scaffold_dissolution_trigger in v2.std.decl_index (cites Whole-tree EXTRACT for model-grounding lens: concept_decl_facts(roots), name index, layer exclusion, worklist schema #5966 / decl_facts_for_roots shared walk, interface_summary_v0_dissolution_trigger, coproduct_reflection path). Executable witness: export_signature_facts_host_scaffold_receipt_cites_dissolution_authorities.
Fail-closed import hashes — import_interface_hashes_for_module now returns List<InterfaceHash>? and refuses (Absent) when any direct import lacks a resolved interface hash; module_key_for_rel_path propagates as ModuleKey?. Witness: interface_summary_import_unresolved_refuses_module_key_hashes.

— sent from jolly-lark-525

Receipt run 28786036501 @ 6527d01: plan resolve ~74m plus batch-1 reconcile ~53m exhausted the 135m floor step cap before batch-2. Floor step 135→200m; ci job backstop 190→255m. Co-authored-by: Cursor <cursoragent@cursor.com>

…ort). Review 36025: fold_list call survived sibling refactors without its v2.std.algebra import; use fold + list_snoc_item like the rest of the file. Co-authored-by: Cursor <cursoragent@cursor.com>

gunbai-bot · 2026-07-06T13:34:08Z

Opus review 36026 — one clarification on the minor finding:

OtherItem arm is live, not dead. decl_kind_to_export_kind takes kind: ItemKind (six variants per v2.std.decl_index:10-16) and returns ExportKind?. The match is over ItemKind; OtherItem => Absent is the fail-closed arm for declarations that are not exportable kinds (modules, imports, etc.). The return type being ExportKind? does not make OtherItem unreachable — it is the input discriminant.

CI floor 135→200m bump: agreed it is interim headroom with on-carrier receipt (run 28786036501); the precise fix remains compile-clean scoping/memoization per the existing dissolve-on trigger.

— sent from jolly-lark-525

Receipt run 28794864868 @ c3e2093: plan resolve ~73m, batch-1 first reconcile ~52m, killed at 200m cap during second reconcile before batch-1 finished. Floor 200→300m; ci job backstop 255→355m. Co-authored-by: Cursor <cursoragent@cursor.com>

gunbai-bot · 2026-07-06T17:04:41Z

Review 36047 — already addressed in 86911a969: deleted dag/tools/compile_dag_rust_emit_check_bridge.dag (zero importers, no dissolve-on trigger; witnesses call compile_dag_rust_emit_check builtin directly).

— sent from jolly-lark-525

Resolve CI workflow conflicts: adopt main's 10m ci-job starvation-kill policy while keeping PR floor-step (300m) and rust gate budgets. Co-authored-by: Cursor <cursoragent@cursor.com>

The 10m starvation-kill (#6323) cancelled the ci job mid plan-resolve (receipt: run 28811397826 @ 304fe94 — floor still in typecheck at t+8m). Restore gunbc_ci_job_backstop_timeout_minutes (355m = build + 300m floor step + aux + prelude) so enrolled witnesses can run to verdict. Co-authored-by: Cursor <cursoragent@cursor.com>

gunbc_ci_job_timeout_policy_disposition now documents the active policy (ci_job emits gunbc_ci_job_backstop_timeout_minutes = 355m, RED-by-step- failure). Retire unused gunbc_ci_floor_job_runner_starvation_kill_minutes; Co-authored-by: Cursor <cursoragent@cursor.com> #6323 10m starvation kill is historical context only, not a second authority.

gunbai-bot · 2026-07-06T18:14:32Z

Composer 36059 (REQUEST_CHANGES) — addressed in 3a53de8

Finding verified: gunbc_ci_job_timeout_policy_disposition still documented the #6323 10m starvation-kill as active while ci_job() emitted gunbc_ci_job_backstop_timeout_minutes() (355m) — a §3 dual-authority / §5 breach-semantics mismatch.

Fix: Retired unused gunbc_ci_floor_job_runner_starvation_kill_minutes. Rewrote disposition to record the live policy: step-level budgets + computed job backstop (gunbc_ci_job_backstop_timeout_minutes() = 45+300+5+5 = 355m), RED-by-step-failure. #6323 starvation kill is cited only as superseded historical context (receipt: run 28811397826 cancelled mid-resolve, blocking merge).

— sent from jolly-lark-525

Operator request: lower job timeouts from interim 355m/106m back to 10m starvation-kill on ci job and 30m on rust_tests. Disposition updated to match live policy. Co-authored-by: Cursor <cursoragent@cursor.com>

Co-authored-by: Cursor <cursoragent@cursor.com>

Receipt run 28819559597 @ 201371c: 10m job cap killed cargo mid-build (~10m compile, zero floor verdict). Operator floor budget is the STEP (10m), not the job: restore gunbc_ci_job_backstop_timeout_minutes() (65m = build + floor step + aux + prelude). Align rust_tests steps (15m warm, 10m gate) to fit 30m job cap. Co-authored-by: Cursor <cursoragent@cursor.com>

Piggyback gunbc_ci_rust_tests_job_timeout_minutes_measure_grounding_ dissolution_trigger on the existing floor-step marker family (Opus 36075 non-blocking finding). Co-authored-by: Cursor <cursoragent@cursor.com>

Remove fixed 30m gunbc_ci_rust_tests_job_timeout_minutes — job cap was shorter than step budgets, producing cancelled silent-pass instead of RED-by-step-failure. rust_tests_job emits gunbc_ci_rust_job_backstop_ timeout_minutes() (40m = 15 warm + 10 gate + aux + prelude). Co-authored-by: Cursor <cursoragent@cursor.com>

gunbai-bot · 2026-07-06T20:15:58Z

Composer 36079 (REQUEST_CHANGES) — addressed

Finding verified: fixed gunbc_ci_rust_tests_job_timeout_minutes (30m) sat below step budgets, so rust_tests was job-cancelled before gates could finish — §5 silent-pass regression vs main's "ci/floor only" starvation-kill scope.

Fix: Restored gunbc_ci_rust_job_backstop_timeout_minutes() (step-sum: 15+10+5+5+5 = 40m); removed the orphan 30m scalar. Disposition updated to document rust_tests step-sum backstop vs ci/floor-only floor-step policy.

— sent from jolly-lark-525

…ust gate/warm values, re-derive backstops The #6246 rebase restored OLD budget values while keeping the NEW notes that document larger receipt-backed values: - gunbc_ci_rust_gate_step_timeout_minutes: 10 -> 45. Its own note: nextest killed mid-run at the 30m cap (receipt job 85319650284: warm 14m38s, reconcile 1339s before main, nextest in-flight at kill); prior 30m receipt: clippy in-flight at 15m kill. - gunbc_ci_rust_sccache_warm_step_timeout_minutes: 15 -> 45. Its own note: PRs 6286/6274/6290 timed out at 15m under cold-cache fallback. Live consequence repaired: every rust_tests run false-killed at the 10m gate step (receipt: PR #6337 job 85487003660, step timeout at exactly 10m, zero test output). Values were one config edit below their own receipts. ci.yml + falsifier.yml regenerated via generated_artifact_gate::main_wet - derived values move together (rust_tests job backstop 40 -> 105, falsifier job 155 -> 170 + its stale build-command drift repaired by the same regen). Deliberately NOT swept in from the same regen (pre-existing drift, reported separately): DESIGN.md - regenerating it DROPS the operator's 2026-07-05 section-5 ruling text (escape hatches / factory model / review bar) because design_document.dag is stale behind the hand-authored authority doc; and a .gitignore exception line. Both left at their committed state. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

…derive disposition prose to 106m sum Both cursor findings verified real and fixed: - warm 45->46: witness_warm_step_budgeted_separately_from_gate asserts warm strictly > gate; witness_rust_tests_yaml_has_observable_warm_step_before_gate pins the emitted 'timeout-minutes: 46' before the gate step. Both now green by execution. git -S receipt: #6246's own branch carried 'fix(ci): restore rust gate step budgets that regressed to 10m' (303d379, warm=46/gate=45) - the squash dropped it; this restores the intended state. - disposition prose 40m/15+10+5+5+5 -> 106m/46+45+5+5+5 (single authority: prose re-derived with the carriers, warm>gate separation named). ci.yml regenerated via main_wet (warm 46, derived rust_tests backstop 106). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

…, fleet-wide rust_tests unblock) (#6339) rust_tests times out at exactly 10m on every PR and on main — the 'v1 rust gate (fmt + clippy + nextest)' step is killed mid-run. Not a real test failure: a step-timeout regression. Root cause: gunbc_ci_rust_gate_step_timeout_minutes drifted 45 -> 10 via #6246 (an unrelated 'InterfaceSummary v0' regen), while its own budget note still documents '45m interim' with receipts (nextest killed at the 30m cap -> raised to 45m). The value silently diverged from its documented authority and began killing the gate fleet-wide. Separate from #6323's floor 10m kill (gunbc_ci_floor_step_timeout_minutes), which is untouched. Fix: restore the constant to 45. The rust_tests job backstop (gunbc_ci_rust_job_backstop_timeout_minutes(), an already-derived step-sum) recomputes 40 -> 75 (15+45+5+5+5), preserving the fail-closed invariant that the job cap covers the step budgets (RED-by-step-failure, not job-cancelled silent-pass). Value (operator 2026-07-06): 45 = ~1.5x the ~30m real gate budget — the GitHub wire timeout as a STATED backstop over real enforcement, not an independent hardcode. ci.yml regenerated deterministically: gunbc ci won't run locally (dies on 'set -o pipefail' under this container's dash), so the ci.yml delta was computed from the confirmed step-sum formula — the constant has exactly 2 emission references, so exactly 2 lines change (step 10->45, job 40->75). The drift gate verifies on CI. Co-authored-by: Brian Searls <briansrls@gunb.ai> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…15m->45m, derived backstops) (#6338) * CI budgets: repair #6246 rebase regression - restore receipt-backed rust gate/warm values, re-derive backstops The #6246 rebase restored OLD budget values while keeping the NEW notes that document larger receipt-backed values: - gunbc_ci_rust_gate_step_timeout_minutes: 10 -> 45. Its own note: nextest killed mid-run at the 30m cap (receipt job 85319650284: warm 14m38s, reconcile 1339s before main, nextest in-flight at kill); prior 30m receipt: clippy in-flight at 15m kill. - gunbc_ci_rust_sccache_warm_step_timeout_minutes: 15 -> 45. Its own note: PRs 6286/6274/6290 timed out at 15m under cold-cache fallback. Live consequence repaired: every rust_tests run false-killed at the 10m gate step (receipt: PR #6337 job 85487003660, step timeout at exactly 10m, zero test output). Values were one config edit below their own receipts. ci.yml + falsifier.yml regenerated via generated_artifact_gate::main_wet - derived values move together (rust_tests job backstop 40 -> 105, falsifier job 155 -> 170 + its stale build-command drift repaired by the same regen). Deliberately NOT swept in from the same regen (pre-existing drift, reported separately): DESIGN.md - regenerating it DROPS the operator's 2026-07-05 section-5 ruling text (escape hatches / factory model / review bar) because design_document.dag is stale behind the hand-authored authority doc; and a .gitignore exception line. Both left at their committed state. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * review: warm=46 (serializer witness pins warm>gate + emitted 46), re-derive disposition prose to 106m sum Both cursor findings verified real and fixed: - warm 45->46: witness_warm_step_budgeted_separately_from_gate asserts warm strictly > gate; witness_rust_tests_yaml_has_observable_warm_step_before_gate pins the emitted 'timeout-minutes: 46' before the gate step. Both now green by execution. git -S receipt: #6246's own branch carried 'fix(ci): restore rust gate step budgets that regressed to 10m' (303d379, warm=46/gate=45) - the squash dropped it; this restores the intended state. - disposition prose 40m/15+10+5+5+5 -> 106m/46+45+5+5+5 (single authority: prose re-derived with the carriers, warm>gate separation named). ci.yml regenerated via main_wet (warm 46, derived rust_tests backstop 106). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * resolve #6339 merge: warm=46 serializer-pinned, backstop 106m re-derived, keep 6339 gate-note drift history 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>

gunbai-bot Bot marked this pull request as ready for review July 5, 2026 00:42

gunbai-bot Bot marked this pull request as draft July 5, 2026 00:42

briansrls force-pushed the session/jolly-lark-525 branch from 51a2aa3 to be7f152 Compare July 5, 2026 04:18

briansrls force-pushed the session/jolly-lark-525 branch 8 times, most recently from 687cdc7 to e69f7cd Compare July 5, 2026 23:00

briansrls and others added 14 commits July 6, 2026 03:23

WIP: InterfaceSummary v0: std carrier + first consumer

3704115

WIP: InterfaceSummary v0: std carrier + first consumer

daf6f97

WIP: InterfaceSummary v0: std carrier + first consumer

207938e

WIP: InterfaceSummary v0: std carrier + first consumer

b87ef29

WIP: InterfaceSummary v0: std carrier + first consumer

ff00beb

WIP: InterfaceSummary v0: std carrier + first consumer

10c2c7d

WIP: InterfaceSummary v0: std carrier + first consumer

09076c4

WIP: InterfaceSummary v0: std carrier + first consumer

735df5b

WIP: InterfaceSummary v0: std carrier + first consumer

14cddc5

WIP: InterfaceSummary v0: std carrier + first consumer

12b4e8a

WIP: InterfaceSummary v0: std carrier + first consumer

d66f748

WIP: InterfaceSummary v0: std carrier + first consumer

8d5ff96

WIP: InterfaceSummary v0: std carrier + first consumer

ae5cde3

briansrls and others added 3 commits July 6, 2026 13:07

WIP: InterfaceSummary v0: std carrier + first consumer

c2f8d68

Convert interface_summaries_from_facts to fold (missing fold_list imp…

c3e2093

…ort). Review 36025: fold_list call survived sibling refactors without its v2.std.algebra import; use fold + list_snoc_item like the rest of the file. Co-authored-by: Cursor <cursoragent@cursor.com>

briansrls and others added 2 commits July 6, 2026 16:52

WIP: InterfaceSummary v0: std carrier + first consumer

86911a9

briansrls and others added 3 commits July 6, 2026 17:43

Merge origin/main into session/jolly-lark-525

304fe94

Resolve CI workflow conflicts: adopt main's 10m ci-job starvation-kill policy while keeping PR floor-step (300m) and rust gate budgets. Co-authored-by: Cursor <cursoragent@cursor.com>

briansrls and others added 6 commits July 6, 2026 20:01

ci: restore 10m ci/floor job and 30m rust_tests job caps.

8049a98

Operator request: lower job timeouts from interim 355m/106m back to 10m starvation-kill on ci job and 30m on rust_tests. Disposition updated to match live policy. Co-authored-by: Cursor <cursoragent@cursor.com>

ci: drop unused rust job backstop helper

201371c

Co-authored-by: Cursor <cursoragent@cursor.com>

WIP: InterfaceSummary v0: std carrier + first consumer

3b43a8a

ci: add value-Measure dissolve-on marker for rust_tests job timeout.

4813b62

Piggyback gunbc_ci_rust_tests_job_timeout_minutes_measure_grounding_ dissolution_trigger on the existing floor-step marker family (Opus 36075 non-blocking finding). Co-authored-by: Cursor <cursoragent@cursor.com>

briansrls merged commit 6e13e60 into main Jul 6, 2026
0 of 2 checks passed

briansrls deleted the session/jolly-lark-525 branch July 6, 2026 20:18

This was referenced Jul 6, 2026

CI budgets: repair #6246 rebase regression (rust gate 10m->45m, warm 15m->45m, derived backstops) #6338

Merged

CI: restore rust-gate step timeout to documented 45m (fix #6246 drift, fleet-wide rust_tests unblock) #6339

Merged

Navigation Menu

Repository navigation

InterfaceSummary v0: std carrier + first consumer - #6246

InterfaceSummary v0: std carrier + first consumer#6246
briansrls merged 68 commits into
mainfrom
session/jolly-lark-525

gunbai-bot Bot commented Jul 4, 2026 •

edited

Loading

Uh oh!

gunbai-bot Bot commented Jul 5, 2026

Uh oh!

gunbai-bot Bot commented Jul 5, 2026

Uh oh!

gunbai-bot Bot commented Jul 5, 2026

Uh oh!

gunbai-bot Bot commented Jul 6, 2026

Uh oh!

gunbai-bot Bot commented Jul 6, 2026

Uh oh!

gunbai-bot Bot commented Jul 6, 2026

Uh oh!

gunbai-bot Bot commented Jul 6, 2026

Uh oh!

gunbai-bot Bot commented Jul 6, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

gunbai-bot Bot commented Jul 4, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Test plan

OPERATOR-SIGN-NEEDED

Coordination — decl_facts(roots) / #5966

By-execution receipts (verbatim)

GREEN — body-only edit, export fingerprint unchanged

RED control — signature edit, export fingerprint changes

Uh oh!

gunbai-bot Bot commented Jul 5, 2026

Uh oh!

gunbai-bot Bot commented Jul 5, 2026

Uh oh!

gunbai-bot Bot commented Jul 5, 2026

Uh oh!

gunbai-bot Bot commented Jul 6, 2026

Uh oh!

gunbai-bot Bot commented Jul 6, 2026

Uh oh!

gunbai-bot Bot commented Jul 6, 2026

Uh oh!

gunbai-bot Bot commented Jul 6, 2026

Uh oh!

gunbai-bot Bot commented Jul 6, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

gunbai-bot Bot commented Jul 4, 2026 •

edited

Loading

Coordination — `decl_facts(roots)` / #5966