Publish DevOpsBench-100 v3.2.6 human-request release

reports/build.json

@@ -84,13 +84,13 @@
     "median": 2,
     "min": 1
   },
-  "prompt_max_pairwise_jaccard": 0.681416,
+  "prompt_max_pairwise_jaccard": 0.673267,
   "prompt_repeated_leading_phrases": 0,
   "prompt_tool_hits": [],
   "prompt_words": {
-    "max": 172,
-    "median": 131,
-    "min": 104
+    "max": 171,
+    "median": 119,
+    "min": 96
   },
   "quality_gates": {
     "all_agent_visible_assets_unique": true,
@@ -176,7 +176,7 @@
     "random_calls": 0,
     "wall_clock_reads": 0
   },
-  "version": "3.2.5",
+  "version": "3.2.6",
   "world_commits": 417,
   "world_documents": 30,
   "world_seeded_rows": 1451,

reports/qualification.json

@@ -30824,5 +30824,5 @@
     "failures": 0,
     "wrong_token_rejections": 100
   },
-  "version": "3.2.5"
+  "version": "3.2.6"
 }

task_files/dob100-001-rca-analytics-egress-blocked/01-employee-request.md

@@ -1 +1 @@
-Analytics connections are refused, not slow. The incident record describes the visible effect, but it does not establish whether the apparent owner, a dependency, or the deployed configuration is responsible. For analytics-worker, the affected business boundary includes reporting freshness, warehouse replica access, scheduled rollups, and downstream decision data. The disputed mechanism is outbound connectivity, DNS reachability, firewall policy, and dependency access. Explain which conclusion survives cross-checking, record it in the existing incident work item, and leave the live evidence that lets the incoming engineer reproduce it; production controls should remain unchanged. That finding is the release gate for the 2026-03-04 recovery of analytics-worker, so the diagnosis and the readiness decision must describe the same operating reality. Can it make that date without taking capacity protected for another team, and is standard supply, paid expedite, or an exception the defensible tradeoff? Leave the dated, costed recommendation where the incoming owner can verify it.
+Analytics connections are refused, not slow. The incident record describes the visible effect, but it does not establish whether the apparent owner, a dependency, or the deployed configuration is responsible. The decision affects reporting freshness, warehouse replica access, scheduled rollups, and downstream decision data. The disputed mechanism is outbound connectivity, DNS reachability, firewall policy, and dependency access. Explain which conclusion survives cross-checking, record it in the existing incident work item, and leave the live evidence that lets the incoming engineer reproduce it; production controls should remain unchanged. This diagnosis will decide whether analytics worker enters recovery on 2026-03-04. Before closing the incident record, give the incoming owner the earliest credible recovery date, the constraint behind it, and whether approved acceleration or an exception would materially help.

task_files/dob100-001-rca-analytics-egress-blocked/51-agent-visible-asset-manifest.json

@@ -1,12 +1,12 @@
 {
   "assets": [
     {
-      "bytes": 1090,
+      "bytes": 927,
       "evidence_role": "request",
       "kind": "md",
       "material": false,
       "path": "01-employee-request.md",
-      "sha256": "999f7fc40a2254944a681b14dbb40f597ae0715592d941515c9d414dc2c0758b",
+      "sha256": "b195a2fd8885a8a5313c742276850d061976855fdb14ef57cf7ffdcf114ae711",
       "source": "employee request"
     },
     {

task_files/dob100-002-rca-analytics-revoked-grant/01-employee-request.md

@@ -1 +1 @@
-Analytics cannot read the replica since the rotation. The incident record describes the visible effect, but it does not establish whether the apparent owner, a dependency, or the deployed configuration is responsible. For analytics-worker, the affected business boundary includes reporting freshness, warehouse replica access, scheduled rollups, and downstream decision data. The disputed mechanism is database authorization, revoked roles, credential scope, and least-privilege access. Resolve the disagreement far enough for the incident lead to act, keep production unchanged, and record both the supported conclusion and any uncertainty the evidence cannot settle. The supported cause controls whether analytics-worker can enter the recovery window on 2026-03-04, making this one incident decision rather than a separate planning exercise. Settle the release promise from the real capacity and calendar, not the target date alone. Compare the feasible paths on timing, spend, authority, and consequence, then leave one verifiable recommendation.
+Analytics cannot read the replica since the rotation. The incident record describes the visible effect, but it does not establish whether the apparent owner, a dependency, or the deployed configuration is responsible. The handoff has to remain sound across reporting freshness, warehouse replica access, scheduled rollups, and downstream decision data. The disputed mechanism is database authorization, revoked roles, credential scope, and least-privilege access. Resolve the disagreement far enough for the incident lead to act, keep production unchanged, and record both the supported conclusion and any uncertainty the evidence cannot settle. Analytics worker is tentatively booked for recovery on 2026-03-04; the evidence here determines whether that promise is credible. Say whether the date holds against current on-call commitments and change windows, what a sanctioned acceleration costs, and which faster route still lacks authority.

task_files/dob100-002-rca-analytics-revoked-grant/51-agent-visible-asset-manifest.json

@@ -1,12 +1,12 @@
 {
   "assets": [
     {
-      "bytes": 1050,
+      "bytes": 943,
       "evidence_role": "request",
       "kind": "md",
       "material": false,
       "path": "01-employee-request.md",
-      "sha256": "48a0b761ee67f0596fa08663beef521450637249f298d022db3437cab11672eb",
+      "sha256": "ae87970643a8b138f88008f7fc165e274447f0c7c60d41483296a434c8514075",
       "source": "employee request"
     },
     {

task_files/dob100-003-rca-checkout-unschedulable-replicas/01-employee-request.md

@@ -1 +1 @@
-Checkout is running at a fraction of capacity. The incident record describes the visible effect, but it does not establish whether the apparent owner, a dependency, or the deployed configuration is responsible. For checkout, the affected business boundary includes cart conversion, order creation, idempotency, and the customer's final purchase step. Give the next shift a defensible finding and make clear which current observations rule out the most tempting alternative explanations. The incident owner will use this conclusion for the 2026-03-04 release of checkout; a diagnosis that cannot support that decision is not finished. Can the authorised choices deliver on time, and which one offers the best timing-cost tradeoff without borrowing another team's reservation? Save the answer and make it independently checkable.
+Checkout is running at a fraction of capacity. The incident record describes the visible effect, but it does not establish whether the apparent owner, a dependency, or the deployed configuration is responsible. The service owner is accountable for cart conversion, order creation, idempotency, and the customer's final purchase step. Give the next shift a defensible finding and make clear which current observations rule out the most tempting alternative explanations. The incident owner is using this finding to set the 2026-03-04 recovery plan for checkout. Turn the finding into a recovery promise the next shift can defend: supported date, schedule variance, cost, approval basis, and the consequence of waiting.

task_files/dob100-003-rca-checkout-unschedulable-replicas/51-agent-visible-asset-manifest.json

@@ -1,12 +1,12 @@
 {
   "assets": [
     {
-      "bytes": 828,
+      "bytes": 718,
       "evidence_role": "request",
       "kind": "md",
       "material": false,
       "path": "01-employee-request.md",
-      "sha256": "b0b198be254c73f606c3835aaab877373dd2124e64f16dc994823f9ea41c1a1e",
+      "sha256": "f53b93f6a923c42937c52f62bebd3a9a74805fb3f6ce6d3d2b8c821ecfb24b95",
       "source": "employee request"
     },
     {

task_files/dob100-004-rca-media-disk-pressure/01-employee-request.md

@@ -1 +1 @@
-Media uploads stalling. The incident record describes the visible effect, but it does not establish whether the apparent owner, a dependency, or the deployed configuration is responsible. For media-service, the affected business boundary includes customer uploads, image transformation, object storage, and CDN delivery behavior. Establish the operative cause instead of restating the alert, keep production unchanged, and record why the present system state supports that conclusion in the existing incident work item. That finding is the release gate for the 2026-03-04 recovery of media-service, so the diagnosis and the readiness decision must describe the same operating reality. Decide whether the current pool, an expedited addition, or a separately approved exception can support the date without displacing protected work. Preserve the evidence behind the chosen completion.
+Media uploads stalling. The incident record describes the visible effect, but it does not establish whether the apparent owner, a dependency, or the deployed configuration is responsible. Current customer and operator exposure includes customer uploads, image transformation, object storage, and CDN delivery behavior. Establish the operative cause instead of restating the alert, keep production unchanged, and record why the present system state supports that conclusion in the existing incident work item. This diagnosis will decide whether media service enters recovery on 2026-03-04. Compare the staffed recovery path with paid acceleration and a separately approved exception, without borrowing capacity protected for another incident.

task_files/dob100-004-rca-media-disk-pressure/51-agent-visible-asset-manifest.json

@@ -1,12 +1,12 @@
 {
   "assets": [
     {
-      "bytes": 884,
+      "bytes": 742,
       "evidence_role": "request",
       "kind": "md",
       "material": false,
       "path": "01-employee-request.md",
-      "sha256": "476a62e6796e684f42bf8c5673e0cfa3e403e9f71f6a4bb3aeacf8c78673e850",
+      "sha256": "14d73239a646549bcae22677dcf76188c3db21d5c93fd3ff54c585d936cb69ea",
       "source": "employee request"
     },
     {

task_files/dob100-005-rca-notifications-node-deadlock/01-employee-request.md

@@ -1 +1 @@
-Notifications are not being delivered. The incident record describes the visible effect, but it does not establish whether the apparent owner, a dependency, or the deployed configuration is responsible. For notifications, the affected business boundary includes receipts, retry policy, delivery templates, and downstream messaging reliability. Separate upstream cause from downstream symptoms and leave a concise, evidence-linked handoff that another engineer can independently challenge. The supported cause controls whether notifications can enter the recovery window on 2026-03-05, making this one incident decision rather than a separate planning exercise. Resolve whether the committed window can be met and what would have to change to improve it. Distinguish an approved acceleration from an attractive but unauthorised shortcut.
+Notifications are not being delivered. The incident record describes the visible effect, but it does not establish whether the apparent owner, a dependency, or the deployed configuration is responsible. The practical blast radius covers receipts, retry policy, delivery templates, and downstream messaging reliability. Separate upstream cause from downstream symptoms and leave a concise, evidence-linked handoff that another engineer can independently challenge. Notifications is tentatively booked for recovery on 2026-03-05; the evidence here determines whether that promise is credible. Before closing the incident record, give the incoming owner the earliest credible recovery date, the constraint behind it, and whether approved acceleration or an exception would materially help.

task_files/dob100-005-rca-notifications-node-deadlock/51-agent-visible-asset-manifest.json

@@ -1,12 +1,12 @@
 {
   "assets": [
     {
-      "bytes": 837,
+      "bytes": 787,
       "evidence_role": "request",
       "kind": "md",
       "material": false,
       "path": "01-employee-request.md",
-      "sha256": "2d3f60cff51a418aadfa5178a39409b0eeb0146165ffb96d6d66665d502d47c0",
+      "sha256": "e157a4a7abb9de945840246979204b1cfe0e6a35d0eda6f5fab3c45df8048ce2",
       "source": "employee request"
     },
     {

task_files/dob100-006-rca-search-unscheduled-reindex/01-employee-request.md

@@ -1 +1 @@
-Search results are stale. The incident record describes the visible effect, but it does not establish whether the apparent owner, a dependency, or the deployed configuration is responsible. For search, the affected business boundary includes index freshness, query relevance, cache behavior, and shopper discovery latency. Determine what the operating evidence supports today and explain why stale records or plausible shortcuts do not change the answer. The incident owner will use this conclusion for the 2026-03-05 release of search; a diagnosis that cannot support that decision is not finished. Which of the available capacity paths is actually supportable once protected load, delivery commitments, change windows, cost, and approval authority are taken seriously? Record the realistic date and any shortfall.
+Search results are stale. The incident record describes the visible effect, but it does not establish whether the apparent owner, a dependency, or the deployed configuration is responsible. For search, a credible answer has to reconcile index freshness, query relevance, cache behavior, and shopper discovery latency. Determine what the operating evidence supports today and explain why stale records or plausible shortcuts do not change the answer. The incident owner is using this finding to set the 2026-03-05 recovery plan for search. Say whether the date holds against current on-call commitments and change windows, what a sanctioned acceleration costs, and which faster route still lacks authority.

task_files/dob100-006-rca-search-unscheduled-reindex/51-agent-visible-asset-manifest.json

@@ -1,12 +1,12 @@
 {
   "assets": [
     {
-      "bytes": 816,
+      "bytes": 706,
       "evidence_role": "request",
       "kind": "md",
       "material": false,
       "path": "01-employee-request.md",
-      "sha256": "4c1066279598e7a279828610766ea06c15c4acc7543d78b9b43383c7f5a39430",
+      "sha256": "c1b23be0b709b4d388692ed1716b74ee5245636808ee627f2f54804b42fc36e6",
       "source": "employee request"
     },
     {

task_files/dob100-007-detect-checkout-errors/01-employee-request.md

@@ -1 +1 @@
-Is checkout violating its SLOs? The current monitoring views do not yet establish whether this is an active objective breach, a stale signal, or a healthy service being blamed for a downstream symptom. For checkout, the affected business boundary includes cart conversion, order creation, idempotency, and the customer's final purchase step. The disputed mechanism is application exceptions, sampled error groups, request failures, and service-level breach. Trace the inconsistency to its authoritative source and record the supported finding while leaving every production control untouched. That finding is the release gate for the 2026-03-05 recovery of checkout, so the diagnosis and the readiness decision must describe the same operating reality. Is the date achievable under current approval, and if not, what is the earliest defensible outcome? Make the tradeoff between waiting, expediting, and requesting an exception explicit for the next owner.
+Is checkout violating its SLOs? The current monitoring views do not yet establish whether this is an active objective breach, a stale signal, or a healthy service being blamed for a downstream symptom. The live operating picture includes cart conversion, order creation, idempotency, and the customer's final purchase step. The disputed mechanism is application exceptions, sampled error groups, request failures, and service-level breach. Trace the inconsistency to its authoritative source and record the supported finding while leaving every production control untouched. This diagnosis will decide whether checkout enters recovery on 2026-03-05. Turn the finding into a recovery promise the next shift can defend: supported date, schedule variance, cost, approval basis, and the consequence of waiting.

task_files/dob100-007-detect-checkout-errors/51-agent-visible-asset-manifest.json

@@ -1,12 +1,12 @@
 {
   "assets": [
     {
-      "bytes": 957,
+      "bytes": 807,
       "evidence_role": "request",
       "kind": "md",
       "material": false,
       "path": "01-employee-request.md",
-      "sha256": "308cdc5210918548d4de3714c69db4ee7bdce68801e900654f829f08c9f48a25",
+      "sha256": "6656661ad1e83a8a3a2dd0b993a7de7caeba1a5b67c7325dbe40203a8f5295a9",
       "source": "employee request"
     },
     {

task_files/dob100-008-detect-inventory/01-employee-request.md

@@ -1 +1 @@
-Determine whether reservation failures are a capacity incident. The current monitoring views do not yet establish whether this is an active objective breach, a stale signal, or a healthy service being blamed for a downstream symptom. For inventory → checkout capability chain, the affected business boundary includes the upstream-to-downstream dependency direction, the contract at each service boundary, ownership, recoverability, and end-to-end health. Reconcile the affected boundary, identify the decisive signal, and record a production-safe conclusion the service owner can use in the existing incident work item. The supported cause controls whether checkout can enter the recovery window on 2026-03-05, making this one incident decision rather than a separate planning exercise. State the earliest credible completion and whether it meets the business window. Include the best approved path and the faster or cheaper alternatives that the evidence or authority rules out.
+Determine whether reservation failures are a capacity incident. The current monitoring views do not yet establish whether this is an active objective breach, a stale signal, or a healthy service being blamed for a downstream symptom. For inventory → checkout capability chain, the operational risk spans the upstream-to-downstream dependency direction, the contract at each service boundary, ownership, recoverability, and end-to-end health. Reconcile the affected boundary, identify the decisive signal, and record a production-safe conclusion the service owner can use in the existing incident work item. Checkout is tentatively booked for recovery on 2026-03-05; the evidence here determines whether that promise is credible. Compare the staffed recovery path with paid acceleration and a separately approved exception, without borrowing capacity protected for another incident.

task_files/dob100-008-detect-inventory/51-agent-visible-asset-manifest.json

@@ -1,12 +1,12 @@
 {
   "assets": [
     {
-      "bytes": 982,
+      "bytes": 884,
       "evidence_role": "request",
       "kind": "md",
       "material": false,
       "path": "01-employee-request.md",
-      "sha256": "3bbaf3e11bd78d3a111568d0298e844860d85e33e7a219dd640b2bc0e18de42f",
+      "sha256": "6e66cfc3ab2c57b68bb3f73a9bc228e7c887c0a5a6834b905f7033ebb74dcf32",
       "source": "employee request"
     },
     {

task_files/dob100-009-rcn-checkout-error-rate/01-employee-request.md

@@ -1 +1 @@
-What is checkout's production error rate? The commerce lead is asking for checkout's current production error rate as a percentage. Sentry and Prometheus give very different numbers. Reconcile the governing definition, time window, identities, and conflicting systems; give the supported answer and explain every material exclusion or unresolved ambiguity. For checkout, the affected business boundary includes cart conversion, order creation, idempotency, and the customer's final purchase step. Use the governing definition and present state to make the decision, then leave the shared record consistent for the next workflow. The team will publish this answer for checkout on 2026-03-05, so data correctness and delivery readiness must agree. What completion can the team responsibly promise, what constraint controls it, and would expediting or seeking broader authority materially improve the result? Record the supported choice in the durable handoff.
+What is checkout's production error rate? The commerce lead is asking for checkout's current production error rate as a percentage. Sentry and Prometheus give very different numbers. Reconcile the governing definition, time window, identities, and conflicting systems; give the supported answer and explain every material exclusion or unresolved ambiguity. The decision must hold across cart conversion, order creation, idempotency, and the customer's final purchase step. Use the governing definition and present state to make the decision, then leave the shared record consistent for the next workflow. The team plans to publish this answer through checkout on 2026-03-05. State when the result can reliably publish after current refreshes, protected workloads, and review capacity, plus what an approved acceleration would cost.

task_files/dob100-009-rcn-checkout-error-rate/51-agent-visible-asset-manifest.json

@@ -1,12 +1,12 @@
 {
   "assets": [
     {
-      "bytes": 958,
+      "bytes": 832,
       "evidence_role": "request",
       "kind": "md",
       "material": false,
       "path": "01-employee-request.md",
-      "sha256": "d186570df5484a4752db197e7f2a4dee7812a00fd80cea94e3f5c09c281e0f65",
+      "sha256": "81bd79bb2f1a1d95642432bf98c8cc08106a1403b92cea3fa8ae7e667c5c4feb",
       "source": "employee request"
     },
     {

task_files/dob100-010-detect-status-page-recurrence/01-employee-request.md

@@ -1 +1 @@
-Customers still cannot check out, and the status page says we fixed it. The current monitoring views do not yet establish whether this is an active objective breach, a stale signal, or a healthy service being blamed for a downstream symptom. For checkout, the affected business boundary includes cart conversion, order creation, idempotency, and the customer's final purchase step. The disputed mechanism is a supposedly resolved incident recurring behind stale customer-status communication. Find the narrowest explanation that fits every material signal and leave the incident record ready for independent verification. That finding is the release gate for the 2026-03-06 recovery of checkout, so the diagnosis and the readiness decision must describe the same operating reality. Give management a decision, not a copied milestone: the supported completion, the binding constraint, the price of acceleration, and the consequence of the option that needs more authority.
+Customers still cannot check out, and the status page says we fixed it. The current monitoring views do not yet establish whether this is an active objective breach, a stale signal, or a healthy service being blamed for a downstream symptom. The next owner needs this settled across cart conversion, order creation, idempotency, and the customer's final purchase step. The disputed mechanism is a supposedly resolved incident recurring behind stale customer-status communication. Find the narrowest explanation that fits every material signal and leave the incident record ready for independent verification. This diagnosis will decide whether checkout enters recovery on 2026-03-06. Say whether the date holds against current on-call commitments and change windows, what a sanctioned acceleration costs, and which faster route still lacks authority.

task_files/dob100-010-detect-status-page-recurrence/51-agent-visible-asset-manifest.json

@@ -1,12 +1,12 @@
 {
   "assets": [
     {
-      "bytes": 972,
+      "bytes": 851,
       "evidence_role": "request",
       "kind": "md",
       "material": false,
       "path": "01-employee-request.md",
-      "sha256": "2febb95fcba0ed7a69f2c6c41a5671175820dc684b7484b47fa82dfb0b70a1c8",
+      "sha256": "87cb0fcb9eec1a6f3ab2a67bc40237a30d8edbac3322068511c42029c195f8ab",
       "source": "employee request"
     },
     {

task_files/dob100-011-detect-storefront-healthy/01-employee-request.md

@@ -1 +1 @@
-Is storefront-web violating its SLOs? The current monitoring views do not yet establish whether this is an active objective breach, a stale signal, or a healthy service being blamed for a downstream symptom. For storefront-web, the affected business boundary includes browser delivery, page rendering, customer navigation, and edge-cache behavior. Test the apparent explanation against the live state, distinguish correlation from cause, keep production unchanged, and record the sourced finding in the existing incident work item. The supported cause controls whether storefront-web can enter the recovery window on 2026-03-06, making this one incident decision rather than a separate planning exercise. Give the owner an honest on-time-or-late answer and the best authorised option, including the cost and operational consequence of the alternatives.
+Is storefront-web violating its SLOs? The current monitoring views do not yet establish whether this is an active objective breach, a stale signal, or a healthy service being blamed for a downstream symptom. The affected workflow reaches browser delivery, page rendering, customer navigation, and edge-cache behavior. Test the apparent explanation against the live state, distinguish correlation from cause, keep production unchanged, and record the sourced finding in the existing incident work item. Storefront web is tentatively booked for recovery on 2026-03-06; the evidence here determines whether that promise is credible. Turn the finding into a recovery promise the next shift can defend: supported date, schedule variance, cost, approval basis, and the consequence of waiting.

task_files/dob100-011-detect-storefront-healthy/51-agent-visible-asset-manifest.json

@@ -1,12 +1,12 @@
 {
   "assets": [
     {
-      "bytes": 853,
+      "bytes": 787,
       "evidence_role": "request",
       "kind": "md",
       "material": false,
       "path": "01-employee-request.md",
-      "sha256": "f173f3c30e826d84e4b1fadc97259cb1167e96698562e3597a0b8ab8ff07b834",
+      "sha256": "7f3122cdfde6ea809a1dea6b4ec3fbf30a08cef8533ec43914e599574b42e70e",
       "source": "employee request"
     },
     {

task_files/dob100-012-localize-analytics-crashloop/01-employee-request.md

@@ -1 +1 @@
-Alarm 9609 — the error tracker shows nothing. The label attached to the alarm may describe where the symptom was noticed rather than the component or runtime condition that actually caused it. For analytics-worker, the affected business boundary includes reporting freshness, warehouse replica access, scheduled rollups, and downstream decision data. The disputed mechanism is container lifecycle, restart pressure, image state, and process termination evidence. Use the current evidence to settle what can be settled, flag what remains ambiguous, and make the reasoning auditable for the next review. The incident owner will use this conclusion for the 2026-03-06 release of analytics-worker; a diagnosis that cannot support that decision is not finished. Choose the lowest-risk supported path that respects protected capacity and the governed release windows. The final handoff should state its date, cost, approval basis, and schedule variance.
+Alarm 9609 — the error tracker shows nothing. The label attached to the alarm may describe where the symptom was noticed rather than the component or runtime condition that actually caused it. This cannot be closed without accounting for reporting freshness, warehouse replica access, scheduled rollups, and downstream decision data. The disputed mechanism is container lifecycle, restart pressure, image state, and process termination evidence. Use the current evidence to settle what can be settled, flag what remains ambiguous, and make the reasoning auditable for the next review. The incident owner is using this finding to set the 2026-03-06 recovery plan for analytics worker. Compare the staffed recovery path with paid acceleration and a separately approved exception, without borrowing capacity protected for another incident.

task_files/dob100-012-localize-analytics-crashloop/51-agent-visible-asset-manifest.json

@@ -1,12 +1,12 @@
 {
   "assets": [
     {
-      "bytes": 950,
+      "bytes": 839,
       "evidence_role": "request",
       "kind": "md",
       "material": false,
       "path": "01-employee-request.md",
-      "sha256": "b72bb4bc102c91e31a0f08f063a309c49ed94e81c9622dd3a7b3fd480eed4d66",
+      "sha256": "c74fd7a7636456d52e36db134a24f66073d51736f95c626fd604fe0f55677cbe",
       "source": "employee request"
     },
     {

task_files/dob100-013-localize-analytics-errors/01-employee-request.md

@@ -1 +1 @@
-Alarm 9609 (analytics-worker errors). The label attached to the alarm may describe where the symptom was noticed rather than the component or runtime condition that actually caused it. For analytics-worker, the affected business boundary includes reporting freshness, warehouse replica access, scheduled rollups, and downstream decision data. The disputed mechanism is application exceptions, sampled error groups, request failures, and service-level breach. Resolve the disagreement far enough for the incident lead to act, keep production unchanged, and record both the supported conclusion and any uncertainty the evidence cannot settle. The incident owner will use this conclusion for the 2026-03-06 release of analytics-worker; a diagnosis that cannot support that decision is not finished. Which of the available capacity paths is actually supportable once protected load, delivery commitments, change windows, cost, and approval authority are taken seriously? Record the realistic date and any shortfall.
+Alarm 9609 (analytics-worker errors). The label attached to the alarm may describe where the symptom was noticed rather than the component or runtime condition that actually caused it. For analytics worker, the operational risk spans reporting freshness, warehouse replica access, scheduled rollups, and downstream decision data. The disputed mechanism is application exceptions, sampled error groups, request failures, and service-level breach. Resolve the disagreement far enough for the incident lead to act, keep production unchanged, and record both the supported conclusion and any uncertainty the evidence cannot settle. The incident owner is using this finding to set the 2026-03-06 recovery plan for analytics worker. Say whether the date holds against current on-call commitments and change windows, what a sanctioned acceleration costs, and which faster route still lacks authority.

task_files/dob100-013-localize-analytics-errors/51-agent-visible-asset-manifest.json

@@ -1,12 +1,12 @@
 {
   "assets": [
     {
-      "bytes": 1012,
+      "bytes": 894,
       "evidence_role": "request",
       "kind": "md",
       "material": false,
       "path": "01-employee-request.md",
-      "sha256": "48fd6b4db53c2f067132c41863d10431cf293e64e0514b58f4bea7b2a37a0cc8",
+      "sha256": "be061d351c2a9a3fe96494d57530fc6e3e9e3fc1cef4b28d140526045382b893",
       "source": "employee request"
     },
     {

task_files/dob100-014-localize-checkout-latency/01-employee-request.md

@@ -1 +1 @@
-Alarm 9610 (checkout latency). The label attached to the alarm may describe where the symptom was noticed rather than the component or runtime condition that actually caused it. For checkout → payments capability chain, the affected business boundary includes the upstream-to-downstream dependency direction, the contract at each service boundary, ownership, recoverability, and end-to-end health. Give the next shift a defensible finding and make clear which current observations rule out the most tempting alternative explanations. That finding is the release gate for the 2026-03-06 recovery of payments, so the diagnosis and the readiness decision must describe the same operating reality. Is the date achievable under current approval, and if not, what is the earliest defensible outcome? Make the tradeoff between waiting, expediting, and requesting an exception explicit for the next owner.
+Alarm 9610 (checkout latency). The label attached to the alarm may describe where the symptom was noticed rather than the component or runtime condition that actually caused it. The decision must hold across the upstream-to-downstream dependency direction, the contract at each service boundary, ownership, recoverability, and end-to-end health. Give the next shift a defensible finding and make clear which current observations rule out the most tempting alternative explanations. This diagnosis will decide whether payments enters recovery on 2026-03-06. Turn the finding into a recovery promise the next shift can defend: supported date, schedule variance, cost, approval basis, and the consequence of waiting.

task_files/dob100-014-localize-checkout-latency/51-agent-visible-asset-manifest.json

@@ -1,12 +1,12 @@
 {
   "assets": [
     {
-      "bytes": 900,
+      "bytes": 714,
       "evidence_role": "request",
       "kind": "md",
       "material": false,
       "path": "01-employee-request.md",
-      "sha256": "0454b02596b2f8e17b74d29fa8b230cbc704a4c7891d559f0f14ae87c4dbf923",
+      "sha256": "f0c9e575d564cec0fcf2d2b4ed1660a4d78085c92f77132b2fee4b18c8e34821",
       "source": "employee request"
     },
     {

task_files/dob100-015-localize-gateway-latency/01-employee-request.md

@@ -1 +1 @@
-Alarm 9604 (api-gateway latency). The label attached to the alarm may describe where the symptom was noticed rather than the component or runtime condition that actually caused it. For api-gateway, the affected business boundary includes edge routing, partner request compatibility, authentication policy, and public latency exposure. Establish the operative cause instead of restating the alert, keep production unchanged, and record why the present system state supports that conclusion in the existing incident work item. The supported cause controls whether api-gateway can enter the recovery window on 2026-03-07, making this one incident decision rather than a separate planning exercise. State the earliest credible completion and whether it meets the business window. Include the best approved path and the faster or cheaper alternatives that the evidence or authority rules out.
+Alarm 9604 (api-gateway latency). The label attached to the alarm may describe where the symptom was noticed rather than the component or runtime condition that actually caused it. The next owner needs this settled across edge routing, partner request compatibility, authentication policy, and public latency exposure. Establish the operative cause instead of restating the alert, keep production unchanged, and record why the present system state supports that conclusion in the existing incident work item. API gateway is tentatively booked for recovery on 2026-03-07; the evidence here determines whether that promise is credible. Compare the staffed recovery path with paid acceleration and a separately approved exception, without borrowing capacity protected for another incident.

task_files/dob100-015-localize-gateway-latency/51-agent-visible-asset-manifest.json

@@ -1,12 +1,12 @@
 {
   "assets": [
     {
-      "bytes": 888,
+      "bytes": 787,
       "evidence_role": "request",
       "kind": "md",
       "material": false,
       "path": "01-employee-request.md",
-      "sha256": "0741ea36ff0f44def131f739c2278e3fafbbc0014408021726c29059c84a0f40",
+      "sha256": "07e3e18a519915a79981ff1dcf88cc50e6c925518a752cb07bf8194e2bc4f336",
       "source": "employee request"
     },
     {

task_files/dob100-016-localize-media-latency/01-employee-request.md

@@ -1 +1 @@
-Alarm 9607 (media-service latency). The label attached to the alarm may describe where the symptom was noticed rather than the component or runtime condition that actually caused it. For media-service, the affected business boundary includes customer uploads, image transformation, object storage, and CDN delivery behavior. Separate upstream cause from downstream symptoms and leave a concise, evidence-linked handoff that another engineer can independently challenge. The incident owner will use this conclusion for the 2026-03-07 release of media-service; a diagnosis that cannot support that decision is not finished. What completion can the team responsibly promise, what constraint controls it, and would expediting or seeking broader authority materially improve the result? Record the supported choice in the durable handoff.
+Alarm 9607 (media-service latency). The label attached to the alarm may describe where the symptom was noticed rather than the component or runtime condition that actually caused it. The affected workflow reaches customer uploads, image transformation, object storage, and CDN delivery behavior. Separate upstream cause from downstream symptoms and leave a concise, evidence-linked handoff that another engineer can independently challenge. The incident owner is using this finding to set the 2026-03-07 recovery plan for media service. Before closing the incident record, give the incoming owner the earliest credible recovery date, the constraint behind it, and whether approved acceleration or an exception would materially help.

task_files/dob100-016-localize-media-latency/51-agent-visible-asset-manifest.json

@@ -1,12 +1,12 @@
 {
   "assets": [
     {
-      "bytes": 834,
+      "bytes": 733,
       "evidence_role": "request",
       "kind": "md",
       "material": false,
       "path": "01-employee-request.md",
-      "sha256": "4a066854b26816bd34bbf325bfe166e6763d3584a85b86000eed142d912cd8c0",
+      "sha256": "725cdd44e3f63ef16544bb010c9933b50fa1626ed7e2e28a9e0868f05a3da1d4",
       "source": "employee request"
     },
     {

task_files/dob100-017-auth-v1-to-v2/01-employee-request.md

@@ -1 +1 @@
-Migrate /v1/auth traffic to /v2/auth and retire v1. Consumers still span the old and replacement contract, so establish actual usage, compatibility, and rollback conditions before completing the traffic transition. For api-gateway → storefront-web capability chain, the affected business boundary includes the upstream-to-downstream dependency direction, the contract at each service boundary, ownership, recoverability, and end-to-end health. The change boundary covers tokens, identity sessions, compatibility headers, and authentication consumers. Choose the supported rollout path from the evidence available today, stop if its control conditions are not met, and document what you observed afterward. This production change and the 2026-03-07 cutover for api-gateway are one release decision. Give management a decision, not a copied milestone: the supported completion, the binding constraint, the price of acceleration, and the consequence of the option that needs more authority.
+Migrate /v1/auth traffic to /v2/auth and retire v1. Consumers still span the old and replacement contract, so establish actual usage, compatibility, and rollback conditions before completing the traffic transition. This cannot be closed without accounting for the upstream-to-downstream dependency direction, the contract at each service boundary, ownership, recoverability, and end-to-end health. The change boundary covers tokens, identity sessions, compatibility headers, and authentication consumers. Choose the supported rollout path from the evidence available today, stop if its control conditions are not met, and document what you observed afterward. The supported change is supposed to take API gateway through its 2026-03-07 release window. Give the release owner an honest completion date and show whether the normal path, paid expedite, or a governed exception best protects customers and rollback safety.

task_files/dob100-017-auth-v1-to-v2/51-agent-visible-asset-manifest.json

@@ -1,12 +1,12 @@
 {
   "assets": [
     {
-      "bytes": 990,
+      "bytes": 919,
       "evidence_role": "request",
       "kind": "md",
       "material": false,
       "path": "01-employee-request.md",
-      "sha256": "2a1fc0bde33c60d4f7f40a21cee44d9c7e940ae83c4c4ae19637897bc1fbaa6c",
+      "sha256": "a4850e9d8af401e213b170fb75bf97f5c6d98e215c069975e4edf5d8396114d5",
       "source": "employee request"
     },
     {

task_files/dob100-018-rca-payments-retry/01-employee-request.md

@@ -1 +1 @@
-Payments error rate. The incident record describes the visible effect, but it does not establish whether the apparent owner, a dependency, or the deployed configuration is responsible. For payments, the affected business boundary includes authorization, settlement, retry safety, and duplicate-charge prevention. Trace the inconsistency to its authoritative source and record the supported finding while leaving every production control untouched. The supported cause controls whether payments can enter the recovery window on 2026-03-07, making this one incident decision rather than a separate planning exercise. Give the owner an honest on-time-or-late answer and the best authorised option, including the cost and operational consequence of the alternatives.
+Payments error rate. The incident record describes the visible effect, but it does not establish whether the apparent owner, a dependency, or the deployed configuration is responsible. The decision affects authorization, settlement, retry safety, and duplicate-charge prevention. Trace the inconsistency to its authoritative source and record the supported finding while leaving every production control untouched. Payments is tentatively booked for recovery on 2026-03-07; the evidence here determines whether that promise is credible. Turn the finding into a recovery promise the next shift can defend: supported date, schedule variance, cost, approval basis, and the consequence of waiting.

task_files/dob100-018-rca-payments-retry/51-agent-visible-asset-manifest.json

@@ -1,12 +1,12 @@
 {
   "assets": [
     {
-      "bytes": 763,
+      "bytes": 694,
       "evidence_role": "request",
       "kind": "md",
       "material": false,
       "path": "01-employee-request.md",
-      "sha256": "476211749d64ff3819e8442995bd4628b1f3c13abd56581fd6a33c27aed89796",
+      "sha256": "0ff657e9ab9bf1e7bf77120e017577f2926a20c08148aa0459f5eab0fc18f767",
       "source": "employee request"
     },
     {

task_files/dob100-019-rca-catalog-n-plus-one/01-employee-request.md

@@ -1 +1 @@
-Catalog pricing latency. The incident record describes the visible effect, but it does not establish whether the apparent owner, a dependency, or the deployed configuration is responsible. For catalog, the affected business boundary includes price publication, product availability, merchandising feeds, and cache-consistent customer display. Reconcile the affected boundary, identify the decisive signal, and record a production-safe conclusion the service owner can use in the existing incident work item. The incident owner will use this conclusion for the 2026-03-07 release of catalog; a diagnosis that cannot support that decision is not finished. Choose the lowest-risk supported path that respects protected capacity and the governed release windows. The final handoff should state its date, cost, approval basis, and schedule variance.
+Catalog pricing latency. The incident record describes the visible effect, but it does not establish whether the apparent owner, a dependency, or the deployed configuration is responsible. The handoff has to remain sound across price publication, product availability, merchandising feeds, and cache-consistent customer display. Reconcile the affected boundary, identify the decisive signal, and record a production-safe conclusion the service owner can use in the existing incident work item. The incident owner is using this finding to set the 2026-03-07 recovery plan for catalog. Compare the staffed recovery path with paid acceleration and a separately approved exception, without borrowing capacity protected for another incident.

task_files/dob100-019-rca-catalog-n-plus-one/51-agent-visible-asset-manifest.json

@@ -1,12 +1,12 @@
 {
   "assets": [
     {
-      "bytes": 845,
+      "bytes": 737,
       "evidence_role": "request",
       "kind": "md",
       "material": false,
       "path": "01-employee-request.md",
-      "sha256": "3ccf6267bcc2c7c70047d04a597ce18a5b6ad4b172af4983dcbaa321ac49bad3",
+      "sha256": "6fbe1cff265eaf6778c497e4e4d5eb3fb2552a6c35b782f7ded7925f2fd39158",
       "source": "employee request"
     },
     {

task_files/dob100-020-rca-notifications-timeout/01-employee-request.md

@@ -1 +1 @@
-Notification delivery failures. The incident record describes the visible effect, but it does not establish whether the apparent owner, a dependency, or the deployed configuration is responsible. For notifications, the affected business boundary includes receipts, retry policy, delivery templates, and downstream messaging reliability. The disputed mechanism is deadline propagation, downstream waiting, retry amplification, and request cancellation. Work through the conflicting records, preserve genuine uncertainty, and state the supported operational conclusion in ordinary language. That finding is the release gate for the 2026-03-08 recovery of notifications, so the diagnosis and the readiness decision must describe the same operating reality. Can it make that date without taking capacity protected for another team, and is standard supply, paid expedite, or an exception the defensible tradeoff? Leave the dated, costed recommendation where the incoming owner can verify it.
+Notification delivery failures. The incident record describes the visible effect, but it does not establish whether the apparent owner, a dependency, or the deployed configuration is responsible. The service owner is accountable for receipts, retry policy, delivery templates, and downstream messaging reliability. The disputed mechanism is deadline propagation, downstream waiting, retry amplification, and request cancellation. Work through the conflicting records, preserve genuine uncertainty, and state the supported operational conclusion in ordinary language. This diagnosis will decide whether notifications enters recovery on 2026-03-08. Before closing the incident record, give the incoming owner the earliest credible recovery date, the constraint behind it, and whether approved acceleration or an exception would materially help.

task_files/dob100-020-rca-notifications-timeout/51-agent-visible-asset-manifest.json

@@ -1,12 +1,12 @@
 {
   "assets": [
     {
-      "bytes": 987,
+      "bytes": 843,
       "evidence_role": "request",
       "kind": "md",
       "material": false,
       "path": "01-employee-request.md",
-      "sha256": "6a60932136743bbe713773f9adf13bf809d27b0bb49255d18aa3633282f9e672",
+      "sha256": "276ae30e4f87b04141c1b489850a484b6c4ed549ea0c2ffe85424c4f9e8b0235",
       "source": "employee request"
     },
     {

task_files/dob100-021-rca-inventory-pool/01-employee-request.md

@@ -1 +1 @@
-Inventory reservation failures. The incident record describes the visible effect, but it does not establish whether the apparent owner, a dependency, or the deployed configuration is responsible. For inventory, the affected business boundary includes stock reservation, warehouse truth, backorder promises, and race-free quantity updates. Find the narrowest explanation that fits every material signal and leave the incident record ready for independent verification. The supported cause controls whether inventory can enter the recovery window on 2026-03-08, making this one incident decision rather than a separate planning exercise. Settle the release promise from the real capacity and calendar, not the target date alone. Compare the feasible paths on timing, spend, authority, and consequence, then leave one verifiable recommendation.
+Inventory reservation failures. The incident record describes the visible effect, but it does not establish whether the apparent owner, a dependency, or the deployed configuration is responsible. Current customer and operator exposure includes stock reservation, warehouse truth, backorder promises, and race-free quantity updates. Find the narrowest explanation that fits every material signal and leave the incident record ready for independent verification. Inventory is tentatively booked for recovery on 2026-03-08; the evidence here determines whether that promise is credible. Say whether the date holds against current on-call commitments and change windows, what a sanctioned acceleration costs, and which faster route still lacks authority.

task_files/dob100-021-rca-inventory-pool/51-agent-visible-asset-manifest.json

@@ -1,12 +1,12 @@
 {
   "assets": [
     {
-      "bytes": 842,
+      "bytes": 751,
       "evidence_role": "request",
       "kind": "md",
       "material": false,
       "path": "01-employee-request.md",
-      "sha256": "25dddfbddcfa0197868adfecaf5305d360cbe53182fc34d16564582c706c2a29",
+      "sha256": "8ef97422c55b04f7d00265148043b5d9741963f5bf0298a7ee9df77b4ec5c837",
       "source": "employee request"
     },
     {

task_files/dob100-022-rca-inventory-migrator-security-context/01-employee-request.md

@@ -1 +1 @@
-Inventory counts disagree with the database. The incident record describes the visible effect, but it does not establish whether the apparent owner, a dependency, or the deployed configuration is responsible. For inventory, the affected business boundary includes stock reservation, warehouse truth, backorder promises, and race-free quantity updates. Test the apparent explanation against the live state, distinguish correlation from cause, keep production unchanged, and record the sourced finding in the existing incident work item. The incident owner will use this conclusion for the 2026-03-08 release of inventory; a diagnosis that cannot support that decision is not finished. Can the authorised choices deliver on time, and which one offers the best timing-cost tradeoff without borrowing another team's reservation? Save the answer and make it independently checkable.
+Inventory counts disagree with the database. The incident record describes the visible effect, but it does not establish whether the apparent owner, a dependency, or the deployed configuration is responsible. The practical blast radius covers stock reservation, warehouse truth, backorder promises, and race-free quantity updates. Test the apparent explanation against the live state, distinguish correlation from cause, keep production unchanged, and record the sourced finding in the existing incident work item. The incident owner is using this finding to set the 2026-03-08 recovery plan for inventory. Turn the finding into a recovery promise the next shift can defend: supported date, schedule variance, cost, approval basis, and the consequence of waiting.

task_files/dob100-022-rca-inventory-migrator-security-context/51-agent-visible-asset-manifest.json

@@ -1,12 +1,12 @@
 {
   "assets": [
     {
-      "bytes": 878,
+      "bytes": 764,
       "evidence_role": "request",
       "kind": "md",
       "material": false,
       "path": "01-employee-request.md",
-      "sha256": "4e3d7efc9a7d044678329eabf62cb8e3e4a8800865bfbb2b87b471ef64feef0c",
+      "sha256": "e42532b07b180d25220ebc8c379cd57f97d0033979bff78ef8d9433f91b3f35a",
       "source": "employee request"
     },
     {

task_files/dob100-023-rca-inventory-missing-role/01-employee-request.md

@@ -1 +1 @@
-The inventory report has never produced output. The incident record describes the visible effect, but it does not establish whether the apparent owner, a dependency, or the deployed configuration is responsible. For inventory, the affected business boundary includes stock reservation, warehouse truth, backorder promises, and race-free quantity updates. Use the current evidence to settle what can be settled, flag what remains ambiguous, and make the reasoning auditable for the next review. That finding is the release gate for the 2026-03-08 recovery of inventory, so the diagnosis and the readiness decision must describe the same operating reality. Decide whether the current pool, an expedited addition, or a separately approved exception can support the date without displacing protected work. Preserve the evidence behind the chosen completion.
+The inventory report has never produced output. The incident record describes the visible effect, but it does not establish whether the apparent owner, a dependency, or the deployed configuration is responsible. For inventory, a credible answer has to reconcile stock reservation, warehouse truth, backorder promises, and race-free quantity updates. Use the current evidence to settle what can be settled, flag what remains ambiguous, and make the reasoning auditable for the next review. This diagnosis will decide whether inventory enters recovery on 2026-03-08. Compare the staffed recovery path with paid acceleration and a separately approved exception, without borrowing capacity protected for another incident.

task_files/dob100-023-rca-inventory-missing-role/51-agent-visible-asset-manifest.json

@@ -1,12 +1,12 @@
 {
   "assets": [
     {
-      "bytes": 854,
+      "bytes": 718,
       "evidence_role": "request",
       "kind": "md",
       "material": false,
       "path": "01-employee-request.md",
-      "sha256": "8621edbe386f15468ed7afd67f2bc32a952d3f172b166adcc524460d4a567fc9",
+      "sha256": "f37bddb26788a20dc4be4b9d75735d50d17e898a04f7cc3940e7f61b56b90eac",
       "source": "employee request"
     },
     {

task_files/dob100-024-attr-three-at-once/01-employee-request.md

@@ -1 +1 @@
-Three alarms in twenty minutes - is this one incident or three? Three customer-impacting alarms arrived within the same twenty-minute window. Determine whether one mechanism explains them or whether they are independent failures, localize each supported cause, and record a production-safe finding that tells the incident lead how many investigations are actually needed. For media-service, api-gateway, and checkout incident set, the affected business boundary includes temporal correlation versus shared causation, the distinct runtime and deployment evidence for each affected service, and whether one or several incident owners are required. Explain which conclusion survives cross-checking, record it in the existing incident work item, and leave the live evidence that lets the incoming engineer reproduce it; production controls should remain unchanged. The supported cause controls whether checkout can enter the recovery window on 2026-03-08, making this one incident decision rather than a separate planning exercise. Resolve whether the committed window can be met and what would have to change to improve it. Distinguish an approved acceleration from an attractive but unauthorised shortcut.
+Three alarms in twenty minutes - is this one incident or three? Three customer-impacting alarms arrived within the same twenty-minute window. Determine whether one mechanism explains them or whether they are independent failures, localize each supported cause, and record a production-safe finding that tells the incident lead how many investigations are actually needed. The live operating picture includes temporal correlation versus shared causation, the distinct runtime and deployment evidence for each affected service, and whether one or several incident owners are required. Explain which conclusion survives cross-checking, record it in the existing incident work item, and leave the live evidence that lets the incoming engineer reproduce it; production controls should remain unchanged. Checkout is tentatively booked for recovery on 2026-03-08; the evidence here determines whether that promise is credible. Before closing the incident record, give the incoming owner the earliest credible recovery date, the constraint behind it, and whether approved acceleration or an exception would materially help.