feat(product): ship transcript-first Braid terminal

.agent/critical-audit/2026-08-12T02-10-41Z/findings.jsonl

@@ -0,0 +1,4 @@
+{"id":1,"severity":"HIGH","file":"scripts/live-required/tangle-sandbox-worker.mjs:212","defect":"Replay projection discarded execution identity from events.","failureScenario":"A replay stream contains the cursor event for execution-1 and a result event for execution-2; the stress proof accepts the foreign result when event identity is dropped.","status":"measured; resolved","evidence":"Independent read-only probe reproduced acceptance before the fix; final `node --test scripts/live-required.test.mjs` reports 19/19, including the foreign-event rejection at scripts/live-required.test.mjs:1259.","fix":"Preserve top-level or data.executionId on replay records and reject any present identity that differs from the requested execution.","verification":"`node --test scripts/live-required.test.mjs` → 19 passed, 0 failed.","costIfShipped":"1 false proof pass per foreign replay occurrence.","savedIfFixed":"1 rejected false proof pass per foreign replay occurrence."}
+{"id":2,"severity":"HIGH","file":"scripts/live-required/tangle-sandbox-stress.mjs:86","defect":"Cancellation replay detection accepted acknowledgements without a terminal session state.","failureScenario":"The target execution returns first cancelled=true and second cancelled=false while sessionStatus remains running; the helper reports cancellation replay success.","status":"measured; resolved","evidence":"Independent read-only probe reproduced acceptance before the fix; final `node --test scripts/live-required.test.mjs` covers running-state rejection at scripts/live-required.test.mjs:973.","fix":"Require the exact target execution ID, the acknowledgement transition, and sessionStatus=cancelled.","verification":"`node --test scripts/live-required.test.mjs` → 19 passed, 0 failed.","costIfShipped":"1 false cancellation proof pass per non-terminal acknowledgement pair.","savedIfFixed":"1 rejected false cancellation proof pass per non-terminal acknowledgement pair."}
+{"id":3,"severity":"MEDIUM","file":"scripts/live-required/tangle-sandbox-stress.mjs:243","defect":"Cleanup inspected only the first sandbox page.","failureScenario":"An exact proof tag appears at offset 100; cleanup reports success after checking a shorter first page and leaves the tagged sandbox undeleted.","status":"measured; resolved","evidence":"Independent read-only probe reproduced first-page confirmation before the fix; final `node --test scripts/live-required.test.mjs` covers page-two deletion at scripts/live-required.test.mjs:1129.","fix":"Page through all sandbox results and recheck all pages after deletion, rejecting repeated IDs.","verification":"`node --test scripts/live-required.test.mjs` → 19 passed, 0 failed.","costIfShipped":"1 leaked proof sandbox and its associated lifetime cost per hidden exact tag.","savedIfFixed":"1 prevented leaked proof sandbox per hidden exact tag."}
+{"id":4,"severity":"MEDIUM","file":"scripts/live-required/tangle-sandbox-stress.mjs:384","defect":"Concurrent admission and cursor waits could omit an emitted admission from failure evidence.","failureScenario":"The dispatcher emits admission, then the watcher emits an error before its cursor; Promise.all rejects before state.admitted is assigned, so the partial artifact loses the admission receipt.","status":"measured; resolved","evidence":"Independent read-only probe reproduced a null admission before the fix; final `node --test scripts/live-required.test.mjs` covers the persisted receipt at scripts/live-required.test.mjs:1073.","fix":"Capture each successful wait into state before joining the two promises.","verification":"`node --test scripts/live-required.test.mjs` → 19 passed, 0 failed.","costIfShipped":"1 missing admission receipt per mid-dispatch failure artifact.","savedIfFixed":"1 retained admission receipt per mid-dispatch failure artifact."}

.agent/critical-audit/2026-08-12T02-10-41Z/manifest.json

@@ -0,0 +1,26 @@
+{
+  "schemaVersion": 1,
+  "scope": "Tangle sandbox stress and worker scripts with deterministic fake seams",
+  "files": 3,
+  "base": "8cce4c6e8aade3b1fc51f558d65a823ac12a5e1a",
+  "head": "8cce4c6e8aade3b1fc51f558d65a823ac12a5e1a",
+  "workingTree": true,
+  "diffSha256": "b8d943482ed98ce1cbf6c0792da8d4de043ea383c2a0de001af57b20a94423d5",
+  "projectType": "Node.js ESM package",
+  "reviewers": [
+    "A: correctness, process lifecycle, cleanup, and failure persistence",
+    "B: architecture, SDK contract, and boundary coverage",
+    "C: independent read-only adversarial review and focused test pass"
+  ],
+  "mode": "current working-tree review with serial local checks",
+  "verdict": "APPROVE",
+  "counts": {
+    "critical": 0,
+    "high": 2,
+    "medium": 2,
+    "low": 0,
+    "resolved": 4,
+    "unresolved": 0,
+    "dropped": 1
+  }
+}

.agent/critical-audit/2026-08-12T02-10-41Z/summary.md

@@ -0,0 +1,38 @@
+# Audit: Tangle sandbox stress and worker scripts — 8cce4c6e8aade3b1fc51f558d65a823ac12a5e1a..8cce4c6e8aade3b1fc51f558d65a823ac12a5e1a — n=3 files, 4 findings
+
+**Verdict:** APPROVE — 4 measured defects resolved · 0 CRITICAL / 2 HIGH / 2 MEDIUM / 0 LOW
+**Worst:** #1 `scripts/live-required/tangle-sandbox-worker.mjs:212` — foreign replay identity could pass · cost if shipped 1 false proof pass per occurrence
+**Next:** stop
+
+## Scope
+
+| Field | Value |
+|---|---|
+| Files | n=3 via explicit target scope |
+| Base..head | `8cce4c6e8aade3b1fc51f558d65a823ac12a5e1a..8cce4c6e8aade3b1fc51f558d65a823ac12a5e1a` |
+| Project type | Node.js ESM package |
+| Reviewers | A,B,C · serial |
+| Not inspected | Runtime adapters and live network behavior; both were outside the requested scope and tests used fakes |
+
+## Findings — 4 of 5, ranked
+
+| # | Sev | file:line | Defect | Failure scenario (input/state → wrong result) | Status | Evidence | Fix | Verification | Cost if shipped | Saved if fixed |
+|---:|---|---|---|---|---|---|---|---|---:|---:|
+| 1 | HIGH | `scripts/live-required/tangle-sandbox-worker.mjs:212` | Replay projection discarded event execution identity. | Cursor event for execution-1 plus result event for execution-2 → foreign result accepted. | measured; resolved | Independent probe; final regression `scripts/live-required.test.mjs:1259` | Preserve event identity and reject foreign IDs. | `node --test scripts/live-required.test.mjs` → 19/19 | 1 false pass/occurrence | 1 false pass rejected/occurrence |
+| 2 | HIGH | `scripts/live-required/tangle-sandbox-stress.mjs:86` | Cancellation proof ignored terminal state. | Matching acknowledgements with `sessionStatus=running` → cancellation proof accepted. | measured; resolved | Independent probe; final regression `scripts/live-required.test.mjs:973` | Require `sessionStatus=cancelled`. | `node --test scripts/live-required.test.mjs` → 19/19 | 1 false pass/occurrence | 1 false pass rejected/occurrence |
+| 3 | MEDIUM | `scripts/live-required/tangle-sandbox-stress.mjs:243` | Cleanup stopped at page one. | Exact tag at offset 100 → sandbox remains while cleanup reports success. | measured; resolved | Independent probe; final regression `scripts/live-required.test.mjs:1129` | Page every result and verify every page after deletion. | `node --test scripts/live-required.test.mjs` → 19/19 | 1 leaked sandbox/hidden tag | 1 leak prevented/hidden tag |
+| 4 | MEDIUM | `scripts/live-required/tangle-sandbox-stress.mjs:384` | Concurrent failure could omit a received admission. | Admission emitted before cursor error → partial artifact has no admission receipt. | measured; resolved | Independent probe; final regression `scripts/live-required.test.mjs:1073` | Store each successful wait before `Promise.all`. | `node --test scripts/live-required.test.mjs` → 19/19 | 1 missing receipt/failure | 1 receipt retained/failure |
+
+1 dropped (an external fake counter changed despite a receipt saying `dispatched=false`; it did not describe a production-visible failure).
+
+## Assumptions & unverified
+
+| Assumption | Finding it would flip | Check that settles it |
+|---|---|---|
+| The SDK keeps `SessionStatus` values aligned with `completed` and `cancelled`. | Terminal-state checks could reject a valid live result. | SDK declaration `types-C7mxmAil.d.ts:3094` and live proof run. |
+| The account list API honors `limit` and `offset`. | Pagination could miss or repeat cleanup records. | SDK declaration `client-BCwkh2kj.d.ts:382` and live cleanup run. |
+
+## Self-gate
+
+9/9 passed — failed: none.
+1 verdict = decision + 1 number · 2 every finding has file:line · 3 concrete failure scenario · 4 status label · 5 evidence is a pointer · 6 cost both sides · 7 fix + verification per row · 8 zero adjectives standing in for counts · 9 185 words outside tables ≤600.

.agent/critical-audit/2026-08-12T03-41-05Z/findings.jsonl

@@ -0,0 +1 @@
+{"findings":[],"dropped":0}

.agent/critical-audit/2026-08-12T03-41-05Z/manifest.json

@@ -0,0 +1,26 @@
+{
+  "schemaVersion": 1,
+  "scope": "provider-neutral retained execution lifecycle and provider-owned identity binding",
+  "files": 11,
+  "base": "8cce4c6e8aade3b1fc51f558d65a823ac12a5e1a",
+  "head": "8cce4c6e8aade3b1fc51f558d65a823ac12a5e1a",
+  "workingTree": true,
+  "diffSha256": "f62722d6b99529f797363a13035430b5a2307e29e1a771b5c34764f46dc21006",
+  "projectType": "Node.js ESM package",
+  "reviewers": [
+    "A: lifecycle correctness, exact identity, cancellation, and resource cleanup",
+    "B: provider-neutral contract, module boundaries, and recovery architecture",
+    "C: independent callsite, fixture, and restart coverage review"
+  ],
+  "mode": "current working-tree review with serial local checks",
+  "verdict": "APPROVE",
+  "counts": {
+    "critical": 0,
+    "high": 0,
+    "medium": 0,
+    "low": 0,
+    "resolved": 0,
+    "unresolved": 0,
+    "dropped": 0
+  }
+}

.agent/critical-audit/2026-08-12T03-41-05Z/summary.md

@@ -0,0 +1,34 @@
+# Audit: provider-neutral retained execution — 8cce4c6e8aade3b1fc51f558d65a823ac12a5e1a..8cce4c6e8aade3b1fc51f558d65a823ac12a5e1a — n=11 files, 0 findings
+
+**Verdict:** APPROVE — no reproducible lifecycle or identity defect · 0 CRITICAL / 0 HIGH / 0 MEDIUM / 0 LOW
+**Worst:** none — no reproducible failure scenario · cost if shipped 0 measured incidents
+**Next:** stop
+
+## Scope
+
+| Field | Value |
+|---|---|
+| Files | n=11 via explicit retained-runtime scope |
+| Base..head | `8cce4c6e8aade3b1fc51f558d65a823ac12a5e1a..8cce4c6e8aade3b1fc51f558d65a823ac12a5e1a` |
+| Project type | Node.js ESM package |
+| Reviewers | A,B,C · serial |
+| Not inspected | Tangle backend and production composition; explicitly outside this change |
+
+## Findings — 0 of 0, ranked
+
+| # | Sev | file:line | Defect | Failure scenario (input/state → wrong result) | Status | Evidence | Fix | Verification | Cost if shipped | Saved if fixed |
+|---:|---|---|---|---|---|---|---|---|---:|---:|
+
+0 dropped (no reproducible failure scenario / no actionable fix).
+
+## Assumptions & unverified
+
+| Assumption | Finding it would flip | Check that settles it |
+|---|---|---|
+| CLI Bridge discovery maps its provider execution ID to the server run lookup key. | A provider implementation could require another discovery coordinate. | The local retained fixture with Braid ID `run/retained-crash-window` passes restart discovery; inspect any future provider adapter contract before adoption. |
+| Tangle recovery will implement `RetainedExecutionDriver.recover` using its persisted exact reference. | A future adapter could still derive cloud environment identity from the local run ID. | Add the Tangle adapter and run the exact-reference recovery test against its provider. |
+
+## Self-gate
+
+9/9 passed — failed: none.
+1 verdict = decision + 1 number · 2 every finding has file:line · 3 concrete failure scenario · 4 status label · 5 evidence is a pointer · 6 cost both sides · 7 fix + verification per row · 8 zero adjectives standing in for counts · 9 <N> words ≤600 outside tables.

.agent/skill-runs.jsonl

@@ -36,3 +36,8 @@
 {"skill":"/ui-test","ts":"2026-08-11T03:04:27Z","project":"braid-current-ux-wt","target":"Braid packed terminal shell and focused workflows","operatorPrompt":"","durationMin":null,"verdict":"PASS","dispatchedTo":"/verify","operatorOverride":null,"transcriptPath":null,"traceDir":null}
 {"skill":"/critical-audit","ts":"2026-08-11T03:04:27Z","project":"braid-current-ux-wt","target":"Braid current UX diff n=32 files","operatorPrompt":"","durationMin":null,"verdict":"APPROVE","dispatchedTo":"/verify","operatorOverride":null,"transcriptPath":null,"traceDir":null}
 {"skill":"/verify","ts":"2026-08-11T03:04:27Z","project":"braid-current-ux-wt","target":"Braid current packed terminal UX and CLI Bridge checks","operatorPrompt":"","durationMin":null,"verdict":"PASS","dispatchedTo":"/ship","operatorOverride":null,"transcriptPath":null,"traceDir":null}
+{"skill":"/harden","ts":"2026-08-12T02:12:32Z","project":"braid-transcript-first-wt","target":"Tangle sandbox stress and worker lifecycle, replay, cancellation, artifact, and cleanup paths n=3 files","operatorPrompt":"","durationMin":null,"verdict":"APPROVE","dispatchedTo":"/critical-audit","operatorOverride":null,"transcriptPath":null,"traceDir":null}
+{"skill":"/critical-audit","ts":"2026-08-12T02:12:33Z","project":"braid-transcript-first-wt","target":"Tangle sandbox stress and worker scripts with deterministic fake seams n=3 files","operatorPrompt":"","durationMin":null,"verdict":"APPROVE","dispatchedTo":"/stop","operatorOverride":null,"transcriptPath":null,"traceDir":null}
+{"skill":"/harden","ts":"2026-08-12T03:42:22Z","project":"braid-transcript-first-wt","target":"provider-neutral retained execution lifecycle, exact provider identity, recovery, cancellation, and event binding n=11 files","operatorPrompt":"","durationMin":null,"verdict":"APPROVE","dispatchedTo":"/critical-audit","operatorOverride":null,"transcriptPath":null,"traceDir":null}
+{"skill":"/critical-audit","ts":"2026-08-12T03:42:23Z","project":"braid-transcript-first-wt","target":"provider-neutral retained execution lifecycle and provider-owned identity binding n=11 files","operatorPrompt":"","durationMin":null,"verdict":"APPROVE","dispatchedTo":"/stop","operatorOverride":null,"transcriptPath":null,"traceDir":null}
+{"skill":"/converge","ts":"2026-08-12T12:53:14Z","project":"braid-transcript-first-wt","target":"Braid full release check performance failure","operatorPrompt":"","durationMin":null,"verdict":"PASS","dispatchedTo":"/review-to-green","operatorOverride":null,"transcriptPath":null,"traceDir":null}