@cbartens/bkit 5.3.0 → 5.4.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (128) hide show
  1. package/.bkit/README.md +3 -2
  2. package/.bkit/core/.agents/docs.md +1 -1
  3. package/.bkit/core/.agents/rules.md +8 -2
  4. package/.bkit/core/.bkit/README.md +2 -1
  5. package/.bkit/core/.bkit/cloud.md +33 -33
  6. package/.bkit/core/.bkit/comparison.md +1 -1
  7. package/.bkit/core/.bkit/install.md +6 -5
  8. package/.bkit/core/.bkit/lite.md +6 -6
  9. package/.bkit/core/.bkit/pipeline.md +4 -3
  10. package/.bkit/core/.bkit/prompts.md +4 -3
  11. package/.bkit/core/.bkit/providers.md +1 -1
  12. package/.bkit/core/.bkit/templates.md +35 -5
  13. package/.bkit/core/.bkit/trackers.md +7 -8
  14. package/.bkit/core/.bkit/workflows.md +106 -86
  15. package/.bkit/core/docs/templates/bkit-audit-accessibility.md +3 -2
  16. package/.bkit/core/docs/templates/bkit-audit-adoption.md +1 -1
  17. package/.bkit/core/docs/templates/bkit-audit-agents.md +1 -1
  18. package/.bkit/core/docs/templates/bkit-audit-aio-seo.md +1 -1
  19. package/.bkit/core/docs/templates/bkit-audit-analytics.md +1 -1
  20. package/.bkit/core/docs/templates/bkit-audit-annotations.md +1 -1
  21. package/.bkit/core/docs/templates/bkit-audit-app-performance.md +1 -1
  22. package/.bkit/core/docs/templates/bkit-audit-app-store.md +1 -1
  23. package/.bkit/core/docs/templates/bkit-audit-architecture.md +1 -1
  24. package/.bkit/core/docs/templates/bkit-audit-ci-performance.md +4 -2
  25. package/.bkit/core/docs/templates/bkit-audit-config.md +3 -2
  26. package/.bkit/core/docs/templates/bkit-audit-cost.md +3 -2
  27. package/.bkit/core/docs/templates/bkit-audit-coverage.md +1 -1
  28. package/.bkit/core/docs/templates/bkit-audit-database.md +3 -2
  29. package/.bkit/core/docs/templates/bkit-audit-dead-code.md +5 -5
  30. package/.bkit/core/docs/templates/bkit-audit-deferrals.md +1 -1
  31. package/.bkit/core/docs/templates/bkit-audit-dependencies.md +14 -13
  32. package/.bkit/core/docs/templates/bkit-audit-documentation.md +1 -1
  33. package/.bkit/core/docs/templates/bkit-audit-duplication.md +2 -2
  34. package/.bkit/core/docs/templates/bkit-audit-failures.md +1 -1
  35. package/.bkit/core/docs/templates/bkit-audit-flakes.md +3 -2
  36. package/.bkit/core/docs/templates/bkit-audit-gates.md +5 -2
  37. package/.bkit/core/docs/templates/bkit-audit-issues.md +3 -2
  38. package/.bkit/core/docs/templates/bkit-audit-llm.md +3 -2
  39. package/.bkit/core/docs/templates/bkit-audit-logs.md +10 -6
  40. package/.bkit/core/docs/templates/bkit-audit-models.md +1 -1
  41. package/.bkit/core/docs/templates/bkit-audit-privacy.md +1 -1
  42. package/.bkit/core/docs/templates/bkit-audit-prompts.md +5 -5
  43. package/.bkit/core/docs/templates/bkit-audit-red-team.md +3 -2
  44. package/.bkit/core/docs/templates/bkit-audit-reliability.md +3 -2
  45. package/.bkit/core/docs/templates/bkit-audit-repo.md +1 -1
  46. package/.bkit/core/docs/templates/bkit-audit-security.md +11 -9
  47. package/.bkit/core/docs/templates/bkit-audit-slop.md +1 -1
  48. package/.bkit/core/docs/templates/bkit-audit-testing.md +6 -5
  49. package/.bkit/core/docs/templates/bkit-audit-usability.md +1 -1
  50. package/.bkit/core/engine/checks/status-holes.ts +0 -1
  51. package/.bkit/core/engine/checks/tips/catalogue.ts +16 -0
  52. package/.bkit/core/engine/checks/tips/index.ts +49 -27
  53. package/.bkit/core/engine/cli.ts +22 -15
  54. package/.bkit/core/engine/commands/comply.ts +25 -16
  55. package/.bkit/core/engine/commands/install/install.ts +11 -4
  56. package/.bkit/core/engine/commands/land.ts +156 -0
  57. package/.bkit/core/engine/commands/lib.ts +10 -8
  58. package/.bkit/core/engine/commands/registry.ts +2 -0
  59. package/.bkit/core/engine/commands/tips.ts +7 -3
  60. package/.bkit/core/engine/commands/workflow/branches.ts +382 -282
  61. package/.bkit/core/engine/commands/workflow/disk.ts +556 -0
  62. package/.bkit/core/engine/commands/workflow/issues.ts +73 -28
  63. package/.bkit/core/engine/commands/workflow/locks.ts +38 -152
  64. package/.bkit/core/engine/commands/workflow/prs.ts +11 -2
  65. package/.bkit/core/engine/commands/workflow/roadmap.ts +48 -53
  66. package/.bkit/core/engine/commands/workflow/runtime.ts +3 -2
  67. package/.bkit/core/engine/commands/workflow/status.ts +33 -11
  68. package/.bkit/core/engine/commands/workflow/steps.ts +24 -15
  69. package/.bkit/core/engine/commands/workflow/workflows.ts +433 -265
  70. package/.bkit/core/engine/commands/workflow/workstreams.ts +3 -2
  71. package/.bkit/core/engine/commands/workflow/worktrees.ts +11 -1
  72. package/.bkit/core/engine/install/modules/core.ts +15 -0
  73. package/.bkit/core/engine/install/placeholders.ts +13 -17
  74. package/.bkit/core/engine/shared/settings.ts +8 -0
  75. package/.bkit/core/engine/shared/timeouts.ts +8 -7
  76. package/.bkit/core/prompts/bkit-audit-generate.md +47 -4
  77. package/.bkit/core/prompts/bkit-utility-update-decisions.md +1 -1
  78. package/.bkit/core/prompts/bkit-utility-update-deferrals.md +1 -1
  79. package/.bkit/core/prompts/bkit-utility-update-documentation.md +1 -1
  80. package/.bkit/core/prompts/bkit-utility-update-failures.md +1 -1
  81. package/.bkit/core/prompts/bkit-utility-update-knowledge.md +1 -1
  82. package/.bkit/core/prompts/bkit-utility-update-roadmap.md +3 -3
  83. package/.bkit/core/workflows/bkit-auto.md +2 -2
  84. package/.bkit/core/workflows/bkit-build.md +7 -7
  85. package/.bkit/core/workflows/bkit-debug.md +11 -11
  86. package/.bkit/core/workflows/bkit-design.md +20 -11
  87. package/.bkit/core/workflows/bkit-plan.md +23 -13
  88. package/.bkit/core/workflows/bkit-ship.md +82 -93
  89. package/.bkit/shared/.claude/bkit-cloud.md +12 -12
  90. package/.bkit/shared/.claude/settings.json +1 -0
  91. package/.bkit/shared/docs/deployments.md +29 -25
  92. package/.bkit/shared/gitignore +1 -2
  93. package/.bkit/templates/.github/ISSUE_TEMPLATE/bkit-change.yml +2 -2
  94. package/.bkit/templates/.github/workflows/bkit-ci.yml +7 -2
  95. package/.bkit/templates/.husky/pre-commit +6 -0
  96. package/.bkit/templates/.husky/pre-push +7 -0
  97. package/.bkit/templates/.lintstagedrc.json +5 -1
  98. package/.bkit/templates/scripts/cleanup/bkit-disk-prune.machine.mjs +197 -0
  99. package/.bkit/templates/scripts/cleanup/bkit-disk-prune.mjs +164 -0
  100. package/CHANGELOG.md +48 -0
  101. package/README.md +97 -96
  102. package/dist/checks/status-holes.js +0 -1
  103. package/dist/checks/tips/catalogue.js +14 -0
  104. package/dist/checks/tips/index.js +44 -28
  105. package/dist/cli.js +22 -16
  106. package/dist/commands/comply.js +20 -13
  107. package/dist/commands/install/install.js +10 -4
  108. package/dist/commands/land.js +129 -0
  109. package/dist/commands/lib.js +9 -6
  110. package/dist/commands/registry.js +2 -0
  111. package/dist/commands/tips.js +7 -3
  112. package/dist/commands/workflow/branches.js +300 -243
  113. package/dist/commands/workflow/disk.js +452 -0
  114. package/dist/commands/workflow/issues.js +63 -27
  115. package/dist/commands/workflow/locks.js +35 -130
  116. package/dist/commands/workflow/prs.js +5 -2
  117. package/dist/commands/workflow/roadmap.js +49 -56
  118. package/dist/commands/workflow/runtime.js +3 -2
  119. package/dist/commands/workflow/status.js +32 -12
  120. package/dist/commands/workflow/steps.js +18 -11
  121. package/dist/commands/workflow/workflows.js +388 -247
  122. package/dist/commands/workflow/workstreams.js +3 -2
  123. package/dist/commands/workflow/worktrees.js +11 -1
  124. package/dist/install/modules/core.js +13 -0
  125. package/dist/install/placeholders.js +13 -19
  126. package/dist/shared/settings.js +5 -0
  127. package/dist/shared/timeouts.js +8 -7
  128. package/package.json +1 -1
@@ -17,7 +17,7 @@
17
17
  > - Any tools mentioned in `Universal Instructions` are examples only; swap for actual tools in final output; drop inapplicable.
18
18
  > - Use dense token-efficient phrasing; no filler / hedging / restating; tables > prose, bullets > paragraphs.
19
19
  > - Read `.agents/docs.md` → load docs for workflow / step / task; skip missing; cite docs inline where required.
20
- > - Every claim cites evidence (`file:line`, or command / query and its output) and tool behind it.
20
+ > - Every claim cites this run's evidence (`file:line`, or command / query and its output) and tool behind it; a previous report's figure, list, or status is a lead, not evidence.
21
21
  > - Read this document once at session start; do not trust session memory; do not re-read on every loop iteration.
22
22
  > - never widen, narrow, or edit the CI accessibility gate or its allowlists; full-rule and platform audits are report-only.
23
23
  > - never record an assistive-technology participant's name, contact, or health detail.
@@ -42,6 +42,7 @@ Target failure: **a gate that stays green while the unchecked surface is where t
42
42
  - WAI-ARIA Authoring Practices Guide; record date read.
43
43
  - Platform accessibility HIG and store nutrition-label criteria (e.g. Apple HIG and Accessibility Nutrition Labels, Android accessibility guidelines and Accessibility Scanner).
44
44
  - National disability-discrimination guidance `Laws` finds applies (e.g. Australia's DDA and AHRC digital-access guidelines, US ADA via DOJ web guidance).
45
+ - bkit block of `docs/design.md` (bkit version from `.bkit/version`): every rule and section
45
46
 
46
47
  ### Scope
47
48
 
@@ -147,7 +148,7 @@ Verify only when applicable.
147
148
  - [ ] **Step 12 — Native:** Native UI → shared controls: name, role, state (e.g. `accessibilityLabel`, `accessibilityRole`, `accessibilityState`). Screen reader on same primary flow. Platform-split files (e.g. RN `.native`/`.web`) not diverged in labelling. Shared component fixed web bug → confirm native twin (consumer refs both ways; text search — language server under-reports). Platform automated audits report-only per primary-flow screen: iOS Accessibility Inspector or `performAccessibilityAudit()` in XCUITest; Android Accessibility Scanner or Espresso Accessibility Test Framework. Text scaling: largest accessibility text size + 200% font max (large-text scaling non-linear, e.g. Android 14+). Content not clipped/overlapped. Search text-scaling opt-outs on content text (e.g. RN `allowFontScaling={false}`, low `maxFontSizeMultiplier`; SwiftUI capped `dynamicTypeSize`; Android text sized in `dp` not `sp`) — finding unless reason recorded. Map to platform nutrition-label criteria (`Project Instructions`); declare only if common tasks completable with that feature; larger text needs ≥200%. Record honest labels; declaring → `bkit-audit-app-store`. **Done when:** `Native` holds label parity, audit rows, scaling, label map, or no native UI in `Checked`.
148
149
  - [ ] **Step 13 — Gate Proof:** Vacuous-assertion class: checker on unmounted subtree passes; unmatched locator can pass assertion. Confirm gate self-checks fail closed — injected failure reported, injected undersize proves rule ran (not merely no violation when alternatives pass isolated node). Each route settle selector resolves after guarded content renders, including post-skeleton mount. Refactored route may measure skeleton again. **Done when:** self-check results in `Gate`, or no gate recorded.
149
150
  - [ ] **Step 14 — Criteria:** Automated rules cover minority of WCAG success criteria; `Manual` + `Additions` = chosen subset. Primary flows web + native: pass/fail/N/A every WCAG 2.2 Level A and AA criterion. Reuse earlier steps (name step); hand-check rest. Each fail = finding. **Done when:** `Criteria` has no empty criterion.
150
- - [ ] **Step 15 — Standards:** Confirm current version per anchor standard; read changes since previous report's version. `Standards Coverage`: row per anchor + per WCAG 2.2 A/AA criterion → checked (name step; reuse `Criteria`) / not applicable (why) / finding. Unapplied recommended practice → adopt (propose step) / reject (why) / not applicable. **Done when:** no empty row.
151
+ - [ ] **Step 15 — Standards:** Confirm current version per anchor standard; read changes since previous report's version. `Standards Coverage`: row per anchor + per WCAG 2.2 A/AA criterion → checked (name step; reuse `Criteria`) / not applicable (why) / finding. Per bkit anchor (doc section, rule, or issue shape): one row → followed (name step or evidence) / not applicable (why) / finding; departure with no recorded reason (e.g. `docs/decisions.jsonl` row, accepted-divergence entry) = finding. Unapplied recommended practice → adopt (propose step) / reject (why) / not applicable. **Done when:** no empty row.
151
152
  - [ ] **Step 16 — Anything Else:** (a) Since previous report commit touching accessibility (routes, shared controls + platform twins (e.g. `.web`/`.native`), palette/tokens, animations, emails, gate rule set/route list/allowlists, checker version) → one-line verdict each. (b) ~30 min bounded: senior specialist flags no step asks; read since previous run (incidents, new `docs/failures.jsonl`, `bkit-audit-logs`, vendor status) — what would no step catch? **Done when:** `Beyond the Steps` holds both passes, or `none`. Class seen on 2 runs → propose step in `Process Improvements`.
152
153
 
153
154
  ### 🔒 Gate 2: Probes Check
@@ -17,7 +17,7 @@
17
17
  > - Any tools mentioned in `Universal Instructions` are examples only; swap for actual tools in final output; drop inapplicable.
18
18
  > - Use dense token-efficient phrasing; no filler / hedging / restating; tables > prose, bullets > paragraphs.
19
19
  > - Read `.agents/docs.md` → load docs for workflow / step / task; skip missing; cite docs inline where required.
20
- > - Every claim cites evidence (`file:line`, or command / query and its output) and tool behind it.
20
+ > - Every claim cites this run's evidence (`file:line`, or command / query and its output) and tool behind it; a previous report's figure, list, or status is a lead, not evidence.
21
21
  > - Read this document once at session start; do not trust session memory; do not re-read on every loop iteration.
22
22
  > - NEVER print or copy a token, key, or secret value; names and presence only.
23
23
  > - never run `bkit install` / `bkit upgrade`; bkit commands = `bkit check`, `bkit doctor` only.
@@ -17,7 +17,7 @@
17
17
  > - Any tools mentioned in `Universal Instructions` are examples only; swap for actual tools in final output; drop inapplicable.
18
18
  > - Use dense token-efficient phrasing; no filler / hedging / restating; tables > prose, bullets > paragraphs.
19
19
  > - Read `.agents/docs.md` → load docs for workflow / step / task; skip missing; cite docs inline where required.
20
- > - Every claim cites evidence (`file:line`, or command / query and its output) and tool behind it.
20
+ > - Every claim cites this run's evidence (`file:line`, or command / query and its output) and tool behind it; a previous report's figure, list, or status is a lead, not evidence.
21
21
  > - Read this document once at session start; do not trust session memory; do not re-read on every loop iteration.
22
22
  > - NEVER print or copy a token, key, or secret value; names, scopes, and presence only.
23
23
  > - never run `bkit install` / `bkit upgrade`.
@@ -17,7 +17,7 @@
17
17
  > - Any tools mentioned in `Universal Instructions` are examples only; swap for actual tools in final output; drop inapplicable.
18
18
  > - Use dense token-efficient phrasing; no filler / hedging / restating; tables > prose, bullets > paragraphs.
19
19
  > - Read `.agents/docs.md` → load docs for workflow / step / task; skip missing; cite docs inline where required.
20
- > - Every claim cites evidence (`file:line`, or command / query and its output) and tool behind it.
20
+ > - Every claim cites this run's evidence (`file:line`, or command / query and its output) and tool behind it; a previous report's figure, list, or status is a lead, not evidence.
21
21
  > - Read this document once at session start; do not trust session memory; do not re-read on every loop iteration.
22
22
  > - never change search-console, webmaster, robots, sitemap, IndexNow, or firewall settings; record the owner's decision.
23
23
  > - Name a personal-data field and its page for `bkit-audit-privacy`; never copy a personal-data value into the report.
@@ -17,7 +17,7 @@
17
17
  > - Any tools mentioned in `Universal Instructions` are examples only; swap for actual tools in final output; drop inapplicable.
18
18
  > - Use dense token-efficient phrasing; no filler / hedging / restating; tables > prose, bullets > paragraphs.
19
19
  > - Read `.agents/docs.md` → load docs for workflow / step / task; skip missing; cite docs inline where required.
20
- > - Every claim cites evidence (`file:line`, or command / query and its output) and tool behind it.
20
+ > - Every claim cites this run's evidence (`file:line`, or command / query and its output) and tool behind it; a previous report's figure, list, or status is a lead, not evidence.
21
21
  > - Read this document once at session start; do not trust session memory; do not re-read on every loop iteration.
22
22
  > - Name the field and where it lives; never copy a personal-data value into the report.
23
23
  > - Vendor consoles are read-only: NEVER publish a container, change a property, or send a live collection hit.
@@ -17,7 +17,7 @@
17
17
  > - Any tools mentioned in `Universal Instructions` are examples only; swap for actual tools in final output; drop inapplicable.
18
18
  > - Use dense token-efficient phrasing; no filler / hedging / restating; tables > prose, bullets > paragraphs.
19
19
  > - Read `.agents/docs.md` → load docs for workflow / step / task; skip missing; cite docs inline where required.
20
- > - Every claim cites evidence (`file:line`, or command / query and its output) and tool behind it.
20
+ > - Every claim cites this run's evidence (`file:line`, or command / query and its output) and tool behind it; a previous report's figure, list, or status is a lead, not evidence.
21
21
  > - Read this document once at session start; do not trust session memory; do not re-read on every loop iteration.
22
22
  > - never edit a workflow, composite action, or other CI config; a recommended change is a finding for a human.
23
23
  > - NEVER dispatch a workflow, push, or otherwise start a run to verify a fix.
@@ -17,7 +17,7 @@
17
17
  > - Any tools mentioned in `Universal Instructions` are examples only; swap for actual tools in final output; drop inapplicable.
18
18
  > - Use dense token-efficient phrasing; no filler / hedging / restating; tables > prose, bullets > paragraphs.
19
19
  > - Read `.agents/docs.md` → load docs for workflow / step / task; skip missing; cite docs inline where required.
20
- > - Every claim cites evidence (`file:line`, or command / query and its output) and tool behind it.
20
+ > - Every claim cites this run's evidence (`file:line`, or command / query and its output) and tool behind it; a previous report's figure, list, or status is a lead, not evidence.
21
21
  > - Read this document once at session start; do not trust session memory; do not re-read on every loop iteration.
22
22
  > - never change sampling rates, dashboards, or project settings to take a measurement; a gap = finding.
23
23
  > - Every figure names environment, window, percentile, and `n`.
@@ -17,7 +17,7 @@
17
17
  > - Any tools mentioned in `Universal Instructions` are examples only; swap for actual tools in final output; drop inapplicable.
18
18
  > - Use dense token-efficient phrasing; no filler / hedging / restating; tables > prose, bullets > paragraphs.
19
19
  > - Read `.agents/docs.md` → load docs for workflow / step / task; skip missing; cite docs inline where required.
20
- > - Every claim cites evidence (`file:line`, or command / query and its output) and tool behind it.
20
+ > - Every claim cites this run's evidence (`file:line`, or command / query and its output) and tool behind it; a previous report's figure, list, or status is a lead, not evidence.
21
21
  > - Read this document once at session start; do not trust session memory; do not re-read on every loop iteration.
22
22
  > - NEVER submit a build, change a listing, accept an agreement, submit a declaration, or create, rotate, or download a credential.
23
23
  > - never copy certificate, key, token, or password material into the report; expiry, holder, and recovery path only.
@@ -17,7 +17,7 @@
17
17
  > - Any tools mentioned in `Universal Instructions` are examples only; swap for actual tools in final output; drop inapplicable.
18
18
  > - Use dense token-efficient phrasing; no filler / hedging / restating; tables > prose, bullets > paragraphs.
19
19
  > - Read `.agents/docs.md` → load docs for workflow / step / task; skip missing; cite docs inline where required.
20
- > - Every claim cites evidence (`file:line`, or command / query and its output) and tool behind it.
20
+ > - Every claim cites this run's evidence (`file:line`, or command / query and its output) and tool behind it; a previous report's figure, list, or status is a lead, not evidence.
21
21
  > - Read this document once at session start; do not trust session memory; do not re-read on every loop iteration.
22
22
  > - A code graph confirms presence; a missing edge proves nothing. Corroborate every load-bearing claim with a language server or text search before it is a finding.
23
23
  > - Build the graph locally and code-only; NEVER send the corpus or credentials to a model backend.
@@ -17,7 +17,7 @@
17
17
  > - Any tools mentioned in `Universal Instructions` are examples only; swap for actual tools in final output; drop inapplicable.
18
18
  > - Use dense token-efficient phrasing; no filler / hedging / restating; tables > prose, bullets > paragraphs.
19
19
  > - Read `.agents/docs.md` → load docs for workflow / step / task; skip missing; cite docs inline where required.
20
- > - Every claim cites evidence (`file:line`, or command / query and its output) and tool behind it.
20
+ > - Every claim cites this run's evidence (`file:line`, or command / query and its output) and tool behind it; a previous report's figure, list, or status is a lead, not evidence.
21
21
  > - Read this document once at session start; do not trust session memory; do not re-read on every loop iteration.
22
22
  > - never dispatch, rerun, or cancel a workflow. A runner-class experiment is named, not run.
23
23
  > - Billable rates are read this run from the billing pages. never copy a previous report's prices.
@@ -45,6 +45,8 @@ CI time gets re-argued from impressions. Two mistakes recur:
45
45
  - Browser-test sharding and report merge (e.g. Playwright sharding and `merge-reports`), at the version in the lockfile.
46
46
  - Concurrency groups and job timeouts (e.g. GitHub Actions `concurrency` and `timeout-minutes`).
47
47
  - Runner billing (e.g. GitHub-hosted and larger runners, a remote builder such as Depot, mobile-build credits such as EAS). Record the date read.
48
+ - bkit block of `docs/deployments.md` (bkit version from `.bkit/version`): § Run-level decision matrix; § One run per branch or pull request; § Parallel test lanes (one lane per tier); § turbo `//` over-triggers `web_affected` on docs-only diffs (turbo monorepos); § One CI run per dev commit (release PR mirrors the `dev` push run)
49
+ - bkit block of `docs/testing.md` (bkit version from `.bkit/version`): § CI verbosity (`[verbose ci]`)
48
50
 
49
51
  ### Scope
50
52
 
@@ -124,7 +126,7 @@ Verify only when applicable.
124
126
  - [ ] **Step 9 — Local:** Time local gate on whole tree, cache forced off — warm or affected-only numbers are meaningless. Record beside CI critical path. **Done when:** `Local` holds forced whole-tree time, command, whether on critical path.
125
127
  - [ ] **Step 10 — Reruns:** Red lane costs full re-run. Read concurrency block before counting. If in-progress runs not cancelled on integration branch, fix-and-repush = two complete runs; if cancelled, count cancelled minutes separately from re-runs — neither is zero. Flakiest lane slowest lane compounds wall clock; minutes here, cause → `bkit-audit-flakes`. Over Step 4 window: runs attempt >1; jobs failed then passed same commit, per lane — both → minutes. Also commits under both push and PR; runs completing after newer run started. **Done when:** `Reruns` holds those minutes and what concurrency block does.
126
128
  - [ ] **Step 11 — Cost:** For Step 4 window: billable minutes per lane × runner class current per-minute price, remote-builder minutes, mobile-build credits — read prices this run. Derive cost per successful merge to integration branch, re-runs included. Spend outside CI minutes → `bkit-audit-cost`. Before recommending runner class change for CPU-bound lane, name one-run experiment on larger/different-architecture runner comparing cost and duration (running it is next step — changes workflow). Every finding states whether it buys latency or cost. **Done when:** `Cost` holds minutes, current prices, cost per merge, experiment name or "none".
127
- - [ ] **Step 12 — Standards:** Confirm current version per anchor standard; read changes since previous report's version. `Standards Coverage`: one row per DORA metric, host performance metric, caching/sharding/concurrency/billing practice in anchor list → checked (name step) / not applicable (why) / finding. Unapplied recommended practice → adopt (propose step) / reject (why) / not applicable. **Done when:** no empty row.
129
+ - [ ] **Step 12 — Standards:** Confirm current version per anchor standard; read changes since previous report's version. `Standards Coverage`: one row per DORA metric, host performance metric, caching/sharding/concurrency/billing practice in anchor list → checked (name step) / not applicable (why) / finding. Per bkit anchor (doc section, rule, or issue shape): one row → followed (name step or evidence) / not applicable (why) / finding; departure with no recorded reason (e.g. `docs/decisions.jsonl` row, accepted-divergence entry) = finding. Unapplied recommended practice → adopt (propose step) / reject (why) / not applicable. **Done when:** no empty row.
128
130
  - [ ] **Step 13 — Anything Else:** (a) Everything changed since previous report's commit touching pipeline (workflows and composite actions, jobs and `needs:`, runner classes, task-runner config and cache keys, shard settings, remote-build profiles, local pre-push gate) → one-line verdict each. (b) Bounded ~30 min pass: what senior CI engineer flags that no step asks; read since previous run (incidents, new `docs/failures.jsonl` rows) — what would no step catch? **Done when:** `Beyond the Steps` holds both passes, or `none`. Class seen on 2 runs → propose step in `Process Improvements`.
129
131
 
130
132
  ### 🔒 Gate 2: Probes Check
@@ -17,7 +17,7 @@
17
17
  > - Any tools mentioned in `Universal Instructions` are examples only; swap for actual tools in final output; drop inapplicable.
18
18
  > - Use dense token-efficient phrasing; no filler / hedging / restating; tables > prose, bullets > paragraphs.
19
19
  > - Read `.agents/docs.md` → load docs for workflow / step / task; skip missing; cite docs inline where required.
20
- > - Every claim cites evidence (`file:line`, or command / query and its output) and tool behind it.
20
+ > - Every claim cites this run's evidence (`file:line`, or command / query and its output) and tool behind it; a previous report's figure, list, or status is a lead, not evidence.
21
21
  > - Read this document once at session start; do not trust session memory; do not re-read on every loop iteration.
22
22
  > - NEVER print or copy a secret value; names, presence, dates, and truncated SHA-256 fingerprints only.
23
23
  > - Dashboards, CLIs, and APIs read-only; NEVER change a setting, secret, or DNS record.
@@ -41,6 +41,7 @@ Code is reviewed; dashboards are not. Dashboard changes (hosting, database, paym
41
41
  - TLS certificate validity schedule (e.g. CA/Browser Forum Baseline Requirements, Ballot SC-081v3).
42
42
  - DNS CAA (RFC 8659); MTA-STS (RFC 8461) and TLS reporting (RFC 8460).
43
43
  - Mailbox-provider sender requirements (e.g. Gmail, Yahoo) and one-click unsubscribe (RFC 8058).
44
+ - bkit block of `docs/deployments.md` (bkit version from `.bkit/version`): § Autonomous phase-close prerequisites; § Post-merge dev CI auto-routing; § Promotion-lag warning (`Check Default Branch Lags Dev`)
44
45
 
45
46
  ### Scope
46
47
 
@@ -117,7 +118,7 @@ Verify only when applicable.
117
118
  - [ ] **Step 9 — Links:** Native app with deep / invite links → web serves `/.well-known/apple-app-site-association` and `/.well-known/assetlinks.json`; each valid and names bundle id and signing identity of current build (`Platform Settings`). Open invite link on device with app (opens app) and without (reaches working web page). Broken file breaks every link silently. **Done when:** `Links` row per file and device case.
118
119
  - [ ] **Step 10 — Deliverability:** Sends email → sending domains vs mailbox-provider sender requirements; record page thresholds. Spam rate (e.g. Google Postmaster Tools) vs bulk-sender limit. One-click unsubscribe (RFC 8058 `List-Unsubscribe-Post` header + visible link) on every non-transactional message; transactional (sign-in links, receipts) exempt. Bounces, complaints, unsubscribes suppressed; nothing re-adds them. Transactional separate from marketing (subdomain / stream) so marketing complaint cannot sink sign-in mail. **Done when:** `Deliverability` row per requirement: threshold · measurement · verdict.
119
120
  - [ ] **Step 11 — Expiry:** Every credential / account with a date: domains, non-auto-renewing certificates, OAuth client secrets, expiring API keys, source-host tokens, build-service tokens, annual paid plans. Store signing credentials and developer memberships → listed by reference to `bkit-audit-app-store`. Per row: date, owner, what breaks. Due ≤60 days = finding; unknown date = finding. Each non-auto-renewing certificate: issuer and days left measured live (e.g. `openssl s_client`, TLS checker), not read from dashboard; cover custom DB domains, inbound mail hosts, API hosts off main host. Hand-renewed certificate = finding whatever its date; propose automation. **Done when:** `Expiry Calendar` sorted by date, every row with a date or an unknown-date finding.
120
- - [ ] **Step 12 — Standards:** Confirm current version per anchor standard; read changes since previous report's version. `Standards Coverage`: one row per Twelve-Factor III and X requirement, CI secure-use recommendation, and platform checklist item → checked (name step) / not applicable (why) / finding. Unapplied recommended practice → adopt (propose step) / reject (why) / not applicable. **Done when:** no empty row.
121
+ - [ ] **Step 12 — Standards:** Confirm current version per anchor standard; read changes since previous report's version. `Standards Coverage`: one row per Twelve-Factor III and X requirement, CI secure-use recommendation, and platform checklist item → checked (name step) / not applicable (why) / finding. Per bkit anchor (doc section, rule, or issue shape): one row → followed (name step or evidence) / not applicable (why) / finding; departure with no recorded reason (e.g. `docs/decisions.jsonl` row, accepted-divergence entry) = finding. Unapplied recommended practice → adopt (propose step) / reject (why) / not applicable. **Done when:** no empty row.
121
122
  - [ ] **Step 13 — Anything Else:** (a) Everything changed since previous report's commit touching config (declared variables and example env files, hosting / DB / mobile config files, workflows, and rulesets, scheduled jobs, DNS records, every `Dashboard Changes` entry) → one-line verdict each; dashboard setting changed that no step lists = finding, propose adding it to `Platform Settings`. (b) Bounded ~30 min pass: what a senior platform engineer flags that no step asks; read what happened since previous run (incidents, `bkit-audit-logs` reports, new `docs/failures.jsonl` rows, vendor status incidents) — what would no step catch? **Done when:** `Beyond the Steps` holds both passes, or `none`. Class seen on 2 runs → propose step in `Process Improvements`.
122
123
 
123
124
  ### 🔒 Gate 2: Probes Check
@@ -17,7 +17,7 @@
17
17
  > - Any tools mentioned in `Universal Instructions` are examples only; swap for actual tools in final output; drop inapplicable.
18
18
  > - Use dense token-efficient phrasing; no filler / hedging / restating; tables > prose, bullets > paragraphs.
19
19
  > - Read `.agents/docs.md` → load docs for workflow / step / task; skip missing; cite docs inline where required.
20
- > - Every claim cites evidence (`file:line`, or command / query and its output) and tool behind it.
20
+ > - Every claim cites this run's evidence (`file:line`, or command / query and its output) and tool behind it; a previous report's figure, list, or status is a lead, not evidence.
21
21
  > - Read this document once at session start; do not trust session memory; do not re-read on every loop iteration.
22
22
  > - never change a plan, cancel a resource, or edit a billing setting.
23
23
  > - NEVER copy an invoice number, payment-card detail, account id, or secret into the report.
@@ -39,6 +39,7 @@
39
39
  - FinOps Framework (e.g. the current edition). Capabilities this audit uses: Allocation, Anomaly Management, Budgeting, Forecasting, Unit Economics, and Rate Optimization, across Public Cloud, SaaS, and AI.
40
40
  - FOCUS (FinOps Open Cost and Usage Specification); record the version.
41
41
  - Vendor spend-control docs (e.g. host spend management, database spend cap, model-gateway guardrails, error-tracker spike protection); record the date read.
42
+ - bkit block of `docs/deployments.md` (bkit version from `.bkit/version`): § Disable GitHub Code Quality billing
42
43
 
43
44
  ### Scope
44
45
 
@@ -117,7 +118,7 @@ Verify only when applicable.
117
118
 
118
119
  Anomalies since previous report: how found (alert or invoice), time to find, cost avoided. Cap that only notifies, omits line items, or pauses production → owner decision on notify/cap/pause with what each would stop; this run does not set it. **Done when:** `Caps` row per usage-billed vendor, and anomaly list.
119
120
 
120
- - [ ] **Step 9 — Standards:** Confirm current version per anchor standard; read changes since previous report's version. `Standards Coverage`: one row per FinOps capability (Allocation, Anomaly Management, Budgeting, Forecasting, Unit Economics, and Rate Optimization) → checked (name step) / not applicable (why) / finding. Score each capability crawl, walk, or run; name next to raise. Unapplied recommended practice → adopt (propose step) / reject (why) / not applicable. **Done when:** no empty row.
121
+ - [ ] **Step 9 — Standards:** Confirm current version per anchor standard; read changes since previous report's version. `Standards Coverage`: one row per FinOps capability (Allocation, Anomaly Management, Budgeting, Forecasting, Unit Economics, and Rate Optimization) → checked (name step) / not applicable (why) / finding. Score each capability crawl, walk, or run; name next to raise. Per bkit anchor (doc section, rule, or issue shape): one row → followed (name step or evidence) / not applicable (why) / finding; departure with no recorded reason (e.g. `docs/decisions.jsonl` row, accepted-divergence entry) = finding. Unapplied recommended practice → adopt (propose step) / reject (why) / not applicable. **Done when:** no empty row.
121
122
  - [ ] **Step 10 — Anything Else:** (a) Everything changed since previous report's commit touching cost (SDKs, outbound hosts, vendors, plans, crons, schedules, fan-outs, model calls, sample rates, environment, seat changes) → one-line verdict each. (b) Bounded ~30 min pass: what senior FinOps flags that no step asks; read since previous run (incidents, log-audit reports under `docs/audits/`, new `docs/failures.jsonl` rows, vendor status incidents) — what would no step catch? Include invoice charges Steps 4–8 do not explain (new line item, fee, tax, FX, overage). **Done when:** `Beyond the Steps` holds both passes, or `none`. Class seen on 2 runs → propose step in `Process Improvements`.
122
123
 
123
124
  ### 🔒 Gate 2: Probes Check
@@ -17,7 +17,7 @@
17
17
  > - Any tools mentioned in `Universal Instructions` are examples only; swap for actual tools in final output; drop inapplicable.
18
18
  > - Use dense token-efficient phrasing; no filler / hedging / restating; tables > prose, bullets > paragraphs.
19
19
  > - Read `.agents/docs.md` → load docs for workflow / step / task; skip missing; cite docs inline where required.
20
- > - Every claim cites evidence (`file:line`, or command / query and its output) and tool behind it.
20
+ > - Every claim cites this run's evidence (`file:line`, or command / query and its output) and tool behind it; a previous report's figure, list, or status is a lead, not evidence.
21
21
  > - Read this document once at session start; do not trust session memory; do not re-read on every loop iteration.
22
22
  > - never edit workflows, floor files, coverage config, or the gate script; the report states the edit.
23
23
  > - never lower a floor, add an exclusion, or add an ignore hint.
@@ -17,7 +17,7 @@
17
17
  > - Any tools mentioned in `Universal Instructions` are examples only; swap for actual tools in final output; drop inapplicable.
18
18
  > - Use dense token-efficient phrasing; no filler / hedging / restating; tables > prose, bullets > paragraphs.
19
19
  > - Read `.agents/docs.md` → load docs for workflow / step / task; skip missing; cite docs inline where required.
20
- > - Every claim cites evidence (`file:line`, or command / query and its output) and tool behind it.
20
+ > - Every claim cites this run's evidence (`file:line`, or command / query and its output) and tool behind it; a previous report's figure, list, or status is a lead, not evidence.
21
21
  > - Read this document once at session start; do not trust session memory; do not re-read on every loop iteration.
22
22
  > - Remote catalogue queries ONLY use `SELECT` (non-relational: read-only reads); writes run locally, or in explicitly rolled-back transactions; ALWAYS check remote unchanged.
23
23
  > - NEVER push schema to remote by hand; NEVER hard-delete; restore target = throwaway, deleted before report finishes.
@@ -41,6 +41,7 @@ Migrations and integration tests exercise queries; neither verifies privileges,
41
41
  - Platform serverless connection guidance.
42
42
  - DB routine-maintenance docs, edition = hosted major (e.g. PostgreSQL Routine Database Maintenance Tasks).
43
43
  - Safe-migration rule set (e.g. Squawk); record version.
44
+ - bkit block of `docs/setup.md` (bkit version from `.bkit/version`): Rules
44
45
 
45
46
  ### Scope
46
47
 
@@ -112,7 +113,7 @@ Verify only when applicable.
112
113
  - [ ] **Step 15 — Vacuum:** Vacuuming engine (e.g. Postgres) → transaction-ID age per DB and largest tables; dead-tuple ratio, last autovacuum / autoanalyze per hot table. Not vacuumed since previous report, or ID age climbing = finding. Other engine → its routine maintenance (e.g. SQLite `VACUUM`, InnoDB purge lag), or skip with evidence. **Done when:** numbers in `Capacity`, or skip recorded.
113
114
  - [ ] **Step 16 — Connections:** Connection limit and pooler size vs peak by application and role. Serverless direct connections → transaction-mode pooler, pool of 1, prepared statements off (or platform equivalent). Idle-in-transaction count and age. Statement and idle-in-transaction timeouts per client / admin role; neither = finding. **Done when:** numbers and each client constructor read in `Capacity`.
114
115
  - [ ] **Step 17 — Restore:** (a) From source, list every outbound side effect a clone fires (scheduled jobs, HTTP-calling triggers / functions, DB webhooks, foreign-data wrappers) and how each stops: logical restore stripping them first / fails closed in clone / unscheduled right after restore (harmless first run only). Reaches real users or payments without failing closed → no binary restore. No plan → finding, do not restore. (b) Restore latest backup to throwaway target; schema matches migrations; time taken and data lost; delete target. (c) Record backup type, retention, point-in-time recovery, RTO, RPO. No written targets = finding. Retention < time to notice silent corruption = finding. (d) Logical backup outside platform account: age; restore notes cover key material backups never contain. None = finding. **Done when:** `Restore` holds side-effect plan and result, or the replacing finding.
115
- - [ ] **Step 18 — Standards:** Confirm current version per anchor standard; read changes since previous report's version. `Standards Coverage`: one row per checklist item, advisor lint, inspection report → checked (name step) / not applicable (why) / finding. Unapplied recommended practice → adopt (propose step) / reject (why) / not applicable. **Done when:** no empty row.
116
+ - [ ] **Step 18 — Standards:** Confirm current version per anchor standard; read changes since previous report's version. `Standards Coverage`: one row per checklist item, advisor lint, inspection report → checked (name step) / not applicable (why) / finding. Per bkit anchor (doc section, rule, or issue shape): one row → followed (name step or evidence) / not applicable (why) / finding; departure with no recorded reason (e.g. `docs/decisions.jsonl` row, accepted-divergence entry) = finding. Unapplied recommended practice → adopt (propose step) / reject (why) / not applicable. **Done when:** no empty row.
116
117
  - [ ] **Step 19 — Anything Else:** (a) Everything changed since previous report's commit touching DB (migrations, tables / columns, functions and grants, policies, extensions, scheduled jobs, buckets, publications, new DB clients) → one-line verdict each. (b) Bounded ~30 min pass: what a senior DBA flags that no step asks; read since previous run (incidents, new `docs/failures.jsonl` rows) — what would no step catch? Include anything catalogue / logs / advisors surfaced unchecked, with query for next run. **Done when:** `Beyond the Steps` holds both passes, or `none`. Class seen on 2 runs → propose step in `Process Improvements`.
117
118
 
118
119
  ### 🔒 Gate 2: Probes Check
@@ -17,7 +17,7 @@
17
17
  > - Any tools mentioned in `Universal Instructions` are examples only; swap for actual tools in final output; drop inapplicable.
18
18
  > - Use dense token-efficient phrasing; no filler / hedging / restating; tables > prose, bullets > paragraphs.
19
19
  > - Read `.agents/docs.md` → load docs for workflow / step / task; skip missing; cite docs inline where required.
20
- > - Every claim cites evidence (`file:line`, or command / query and its output) and tool behind it.
20
+ > - Every claim cites this run's evidence (`file:line`, or command / query and its output) and tool behind it; a previous report's figure, list, or status is a lead, not evidence.
21
21
  > - Read this document once at session start; do not trust session memory; do not re-read on every loop iteration.
22
22
  > - NEVER delete code, files, routes, schema objects, or dependencies. The report names the destination; applying it is the next step.
23
23
  > - A detector proves use, never absence. Dead only when every method that would have found a use found none, and the file was opened. never write a finding from tool output alone.
@@ -108,8 +108,8 @@ Verify only when applicable.
108
108
  ### Probe Steps
109
109
 
110
110
  - [ ] **Step 1 — Previous Run:** Audited commit = this report's `bkit:anchor` `baseSha`. Read newest prior report for this slug under `docs/audits/`; its commit = its `bkit:anchor` `baseSha`, else the HEAD its `Previous Run` recorded. Count findings fixed / still open; age of oldest open. Open after 2 runs → re-raise with age. **Done when:** closure rate in `Previous Run`, or run marked baseline.
111
- - [ ] **Step 2 — Detectors:** Compiled language → scanner sees module exports only; dead branch/parameter inside live function → compiler + linter. Read resolved config per workspace, not base: unused-local, unused-parameter, no-unreachable-code; linter unused-vars, unreachable, always-true/false (e.g. `@typescript-eslint/no-unnecessary-condition`). Workspace override or file disable switches one off silently. Detector off = finding → `bkit-audit-gates` (gate candidate). **Done when:** `Detectors` lists each config and resolved value.
112
- - [ ] **Step 3 — Reachability:** Ad hoc scanner, same resolved version both runs; record version. Do not add it as a dependency, and do not wire a CI gate this run. Report baseline-config readiness. Gate before `False Positives` exclusions encoded → structural noise → disabled. Do not encode test-only code as ignore entries or negated project patterns — plugins add test files as entries; production mode separates. Check scanner stderr before trusting: plugin load failure drops entry class → false unused files. Signal: files, exports, types, duplicates. Dependency categories → `Dependencies` pass, not mixed. Same version in production mode (or strict: implies production + each workspace uses only its own deps). Export used default run, unused production → test/config-only class. Report both runs side by side. **Done when:** version, both runs, plugin-load errors in `Reachability`.
111
+ - [ ] **Step 2 — Detectors:** Compiled language → scanner sees module exports only; dead branch/parameter inside live function → compiler + linter. Read resolved config per workspace, not base: unused-local, unused-parameter, no-unreachable-code; linter unused-vars, unreachable, always-true/false (e.g. `@typescript-eslint/no-unnecessary-condition`). Workspace override or file disable switches one off silently. Detector off = finding → `bkit-audit-gates` (gate candidate). Linter cannot run on project's compiler version (e.g. peer dependency range excludes it) → same detector-off finding; its rules → `Not Checked`. **Done when:** `Detectors` lists each config and resolved value.
112
+ - [ ] **Step 3 — Reachability:** Ad hoc scanner, same resolved version both runs; record version. Do not add it as a dependency, and do not wire a CI gate this run. Report baseline-config readiness. First readiness check: scanner ignores exports used inside their own file (e.g. knip `ignoreExportsUsedInFile`); without it export signal = noise → run with and without, report both counts. Gate before `False Positives` exclusions encoded → structural noise → disabled. Do not encode test-only code as ignore entries or negated project patterns — plugins add test files as entries; production mode separates. Check scanner stderr before trusting: plugin load failure drops entry class → false unused files. Signal: files, exports, types, duplicates. Dependency categories → `Dependencies` pass, not mixed. Same version in production mode (or strict: implies production + each workspace uses only its own deps). Export used default run, unused production → test/config-only class. Report both runs side by side. Scanner has no class-member check (e.g. knip 6: no `classMembers` issue type) and project has classes → gap in `Not Checked`; fallback = language-server references per public member. **Done when:** version, both runs, with/without in-file-use counts, plugin-load errors in `Reachability`.
113
113
  - [ ] **Step 4 — Dependencies:** Scanner dependency mode per workspace, categories the scanner offers (e.g. knip: unused, unlisted, unresolved, unlisted binaries, catalog entries unreferenced). Before unused verdict → search name across manifest scripts, config, CI. Config-only consumers (e.g. linter, formatter, CSS presets, bundler plugins, `@types/*`) = usual false hits. Unused = install/supply-chain surface; unlisted = hoisted from another workspace. Unlisted or advisory → `bkit-audit-dependencies`. **Done when:** `Dependencies` row per candidate: name · class · config-or-CI reader · verdict.
114
114
  - [ ] **Step 5 — Env Vars:** Text search only — env names are string literals; language server cannot see them. Every referenced env vs declared (examples, task runner, typed env schema — sources in `Project Instructions`). Referenced, never declared → control cannot switch on. Triage: platform-injected = fine; optional + safe default = fine (read default); gates control (absence disables limiter, auth, feature) = finding. Third group → path never executed in any environment. Omitted declaration source → `Process Improvements` gap. Gating var → `docs/failures.jsonl` via `@bkit-utility-update-failures`, and `bkit-audit-duplication`. **Done when:** `Env Vars` row per referenced-undeclared: group · absence effect · verdict.
115
115
  - [ ] **Step 6 — Config:** Deadness no scanner reports. (a) Feature flags stuck on/off; hardcoded or same every environment → branch never runs. Per flag: every read + value per environment; all same → other branch dead = finding. Age flags (`git log -S`); >90 days, rollout complete = finding even if environments differ. (b) Config keys no reader, or name removed object — search key; language server reads typed source only. (c) Default replaced in every environment by later layer = lying docs. Dead control → `docs/failures.jsonl` like `Env Vars`, not tidiness. **Done when:** `Config` row per flag/unread key: value per environment · age · verdict.
@@ -118,8 +118,8 @@ Verify only when applicable.
118
118
  - [ ] **Step 9 — Fetches:** Clients call API paths with raw fetch → the mirror of `Routes`. Route-side asks whether a route has a caller. This asks whether a call has a route. Neither finds the other's class. Compiler-checked RPC clients are out of scope (a missing route fails typecheck). Raw fetch is the whole exposure: a URL inside a template literal, so lint, typecheck, and the reachability scanner all pass on a path that 404s. Two routing layers (where the stack has them) → model both. The outer framework decides whether a handler runs (static directory vs catch-all that forwards a subtree). The inner router then routes within what was forwarded. A route correctly mounted on the inner router and unreachable from the outer layer is its own class (`stranded`). Checking either layer alone says the route is fine. Read a finding as two questions: the path has no handler, and why nobody noticed. The answer is at the call site — `if (res.ok)` skips the block, `if (!res.ok) return success` fails open. A 404 is a successful HTTP response, so nothing throws. The report names the guard and the route. Confirm against the deployment (status code of the path beside a known-live path) before writing it up. Two false positives, do not reintroduce them: a path in a doc comment is not a call; a prefix test (`startsWith`) is not a request. The dangerous failure is silent: an extractor that stops at `?` never reports the URL, so it is not seen. When changing extraction, the fetched-path count must go up, not merely the findings list get shorter. No project script → extract literals and match routes by hand; record the gap in `Process Improvements`. **Done when:** `Fetches` row per unreachable call and per stranded route: path · call-site guard · deployment status · verdict.
119
119
  - [ ] **Step 10 — Schema:** SQL migrations or hosted DB — only step here. Unused column/orphan SQL function still costs migrations + access review. From the schema source (migrations, schema file, or generated types): per table, column, view, SQL function, access policy (e.g. RLS) → search code (table calls, columns, RPC), other SQL (triggers, policies, functions), deployed serverless functions (e.g. edge functions). Zero refs all three = candidate. Runtime on deployed DB: table no scans/writes since stats reset (record reset time); inspection-CLI table/index stats; unused-index advisor. Unused indexes → also `bkit-audit-database` — name, do not duplicate capacity finding. Thin/pre-launch traffic → zero-use = static candidate, never proof alone. **Done when:** `Schema` row per candidate: object · static refs · runtime counts · reset time · verdict.
120
120
  - [ ] **Step 11 — Events:** Jobs, crons, event bus — mirror of `Fetches` for background work. Event names = string literals; scanner + language server blind. List every send, registered function + trigger/cron, every cron + target. Sent event → registered function; registered trigger → sender or cron; cron → existing target. No trigger = dead function; no function = unfireable control → route like `Env Vars`. **Done when:** `Events` row per name: sender · function · cron target · verdict.
121
- - [ ] **Step 12 — Strings:** (a) Message catalogue — diff keys vs reads each app/shared UI both ways (`i18n-check` or search script). No reader = dead; read missing = renders key name; language server often blind → search. Missing catalogue → `bkit-audit-usability`. (b) Static assets — list shipped (incl. email templates); search basename in code, config, CSS; scanner skips assets; native bundle ships unused. (c) Scripts — per manifest script + scripts tree file → search name in CI, hooks, task runner, docs, other scripts; no invoker = candidate; doc-only = candidate + reader to update. **Done when:** `Strings` row per unused key, missing key, unread asset, uninvoked script.
122
- - [ ] **Step 13 — False Positives:** Filter these before reading anything as a finding. They are structural and recur every run. Ambient module augmentation: zero importers by design; applied through the compiler include. Never dead. Platform file splits (e.g. RN `.ios` / `.android` / `.web`): one specifier resolves to a per-platform file the scanner cannot follow, so unresolved entries are noise. The filter hides the opposite, so check once per run: for each platform pair, the platform that would load that variant imports the parent. A single-platform component carrying the other platform's file is dead and invisible, because both resolve through the same specifier. Files referenced from config, not code: check the referencing config before believing unused (app config, CI, test-runner project dependencies, agent hooks, deployed functions, an explicit worker registry). Test-only exports: only importer is a test = a seam, not dead code. Report separately from unreferenced code; never delete. A default export a preview server requires while code imports the named one. Mock-factory keys: no scanner and no language server sees them; a key inside a mock factory is a real use. **Done when:** `Candidates` names each class and what it removed.
121
+ - [ ] **Step 12 — Strings:** (a) Message catalogue — diff keys vs reads each app/shared UI both ways (`i18n-check` or search script). No reader = dead; read missing = renders key name; language server often blind → search. Missing catalogue → `bkit-audit-usability`. (b) Static assets — list shipped (incl. email templates); search basename in code, config, CSS; scanner skips assets; native bundle ships unused. (c) Scripts — per manifest script + scripts tree file → search name in CI, hooks, task runner, docs, other scripts, plus its stem and every process-spawn site (e.g. `spawn`, `exec`, a directory listing filtered by extension) — runtime-built path misses name search; no invoker = candidate; doc-only = candidate + reader to update. **Done when:** `Strings` row per unused key, missing key, unread asset, uninvoked script.
122
+ - [ ] **Step 13 — False Positives:** Filter these before reading anything as a finding. They are structural and recur every run. Ambient module augmentation: zero importers by design; applied through the compiler include. Never dead. Platform file splits (e.g. RN `.ios` / `.android` / `.web`): one specifier resolves to a per-platform file the scanner cannot follow, so unresolved entries are noise. The filter hides the opposite, so check once per run: for each platform pair, the platform that would load that variant imports the parent. A single-platform component carrying the other platform's file is dead and invisible, because both resolve through the same specifier. Files referenced from config, not code: check the referencing config before believing unused (app config, CI, test-runner project dependencies, agent hooks, deployed functions, an explicit worker registry). Test-only exports: only importer is a test = a seam, not dead code. Report separately from unreferenced code; never delete. A drift test is not a reader: generated artefact referenced only by its generator and a drift test (no runtime reader) = finding, not seam. A default export a preview server requires while code imports the named one. Mock-factory keys: no scanner and no language server sees them; a key inside a mock factory is a real use. **Done when:** `Candidates` names each class and what it removed.
123
123
  - [ ] **Step 14 — Eliminate:** Language server before opening files. One reference found means the candidate is not dead, and that verdict is sound even though the converse is not. Per surviving unused-export candidate: find the symbol, then find references. Any hit ends it. Empty → fall through to text search, and re-run the empty language-server query once more before recording it. Record both measurements. Pace reference queries: back-to-back calls collapse to zero, and a partial answer looks exactly like a dead symbol. Empty is inconclusive, not absence. Text search over-reports (comments, same-named symbols in other modules); the language server under-reports. Agreement is the signal. Run search in addition to the language server, never instead of it. An unused export needs zero matches repo-wide other than its own definition: search, subtract the defining file, ignore prose comments. An unused file is not reached from JSON, YAML, or other config no typed tool reads. An env-var finding is a control, not an optional setting with a working default. A test id asserted in end-to-end tests exists on a component. No finding without opening the file. Cite `file:line` and the method. Non-typed surfaces (markdown, ledgers, YAML, SQL, shell, string literals) are search, not the language server. **Done when:** every survivor in `Candidates` has `file:line`, method, and the zero-reference re-run.
124
124
  - [ ] **Step 15 — Runtime:** Production request/job/query stats → static says unreferenced; production says ran. Rank survivors: recent window of per-route counts/logs, job runs, SQL function query stats. "Zero executions" ≠ "zero static references". Thin/pre-launch → signal, never proof; uncalled route ≠ dead. **Done when:** `Runtime` holds window, source, zero-execution rows, or skip.
125
125
  - [ ] **Step 16 — Standards:** Confirm current version per anchor standard; read changes since previous report's version. `Standards Coverage`: row per anchor + scanner issue type → checked (name step) / not applicable (why) / finding. Unapplied recommended practice → adopt (propose step) / reject (why) / not applicable. **Done when:** no empty row.
@@ -17,7 +17,7 @@
17
17
  > - Any tools mentioned in `Universal Instructions` are examples only; swap for actual tools in final output; drop inapplicable.
18
18
  > - Use dense token-efficient phrasing; no filler / hedging / restating; tables > prose, bullets > paragraphs.
19
19
  > - Read `.agents/docs.md` → load docs for workflow / step / task; skip missing; cite docs inline where required.
20
- > - Every claim cites evidence (`file:line`, or command / query and its output) and tool behind it.
20
+ > - Every claim cites this run's evidence (`file:line`, or command / query and its output) and tool behind it; a previous report's figure, list, or status is a lead, not evidence.
21
21
  > - Read this document once at session start; do not trust session memory; do not re-read on every loop iteration.
22
22
  > - NEVER append or edit a ledger line. Correction and close rows live in `Expected` only. `ts` is minted at append (`.bkit/jsonl.md` § Write Idiom), never written in this report.
23
23
  > - A winning row whose `ts` is in the future outranks every correction. Name its `key`. Do not invent a later `ts`.
@@ -17,7 +17,7 @@
17
17
  > - Any tools mentioned in `Universal Instructions` are examples only; swap for actual tools in final output; drop inapplicable.
18
18
  > - Use dense token-efficient phrasing; no filler / hedging / restating; tables > prose, bullets > paragraphs.
19
19
  > - Read `.agents/docs.md` → load docs for workflow / step / task; skip missing; cite docs inline where required.
20
- > - Every claim cites evidence (`file:line`, or command / query and its output) and tool behind it.
20
+ > - Every claim cites this run's evidence (`file:line`, or command / query and its output) and tool behind it; a previous report's figure, list, or status is a lead, not evidence.
21
21
  > - Read this document once at session start; do not trust session memory; do not re-read on every loop iteration.
22
22
  > - NEVER push, merge, label, close, or comment on a PR. NEVER commit onto the integration branch. Throwaway worktree = evidence; remove before report finishes.
23
23
  > - never resolve a lockfile conflict as text, and never with `-X ours` / `-X theirs`. NEVER run a recursive update that rewrites ranges the PRs did not name.
@@ -45,6 +45,7 @@ Every dependency PR rewrites the lockfile. Unbatched → PRs stale against each
45
45
  - Runtime end-of-life catalogue (e.g. endoflife.date); record the date read.
46
46
  - SPDX License List; record version.
47
47
  - Install-script supply-chain attacks (e.g. Shai-Hulud, September 2025, and Shai-Hulud 2.0, November 2025).
48
+ - bkit block of `docs/deployments.md` (bkit version from `.bkit/version`): § Dependency updaters; § Manual CVE handling (when an alert fires)
48
49
 
49
50
  ### Scope
50
51
 
@@ -81,9 +82,9 @@ One per PR. Class (A, B, or C) = evidence; below = outcomes. DECISION wins over
81
82
 
82
83
  Batch verdict, in `Summary`, not per PR. The landing push's CI is the next step's result, not this run's.
83
84
 
84
- - **PASS:** drift check and every local verification command exited 0.
85
+ - **PASS:** drift check and every local verification command exited 0, baseline reds (Step 10) aside.
85
86
  - **CONCERNS:** a PR was dropped during bisection.
86
- - **FAIL:** drift, or a red no single PR explains.
87
+ - **FAIL:** drift, or a red no single PR explains and the audited commit does not show.
87
88
 
88
89
  ### Pre-Flight / Dependencies
89
90
 
@@ -98,7 +99,7 @@ Always verify.
98
99
  - [ ] **Permissions:** Before Step 1 — one cheap read-only call per tool this run uses (every item below, every rule under `Project Instructions` `Permissions`), e.g. `gh run list --limit 1`, one MCP read, one anchor-standard fetch. Blocked / denied (here or in any step) → record call verbatim and narrow allow rule clearing it (e.g. `Bash(gh run list *)`) in `Not Checked`, `Process Improvements`, and `Project Instructions` `Permissions`; run every step the remaining tools allow; work needing blocked call → `Not Checked`. never stop the run for a blocked call; never retry or reword a blocked call to get past it. Call that ran but the API refused (e.g. HTTP 403 from an endpoint needing another token type, such as GitHub `GET /user/installations`) → not a block; `Not Checked` only, no allow rule.
99
100
  - [ ] **Git Host:** Steps 2, 3, 11, 12 — open PRs by head-ref prefix, CI runs, behind-count, e.g. `gh`, GitLab API. Fallback: none; PR steps `Not Checked`, batch verdict `FAIL`.
100
101
  - [ ] **Package Manager:** Steps 5, 8–10 — install, lockfile, registry metadata, e.g. pnpm, npm. Fallback: none; `Batch` and `Verification` `Not Checked`, batch verdict `FAIL`.
101
- - [ ] **Clean Worktree:** Steps 8–10 — detached worktree off the integration branch, removed before the report finishes. Fallback: none; do not build the batch in the main checkout.
102
+ - [ ] **Clean Worktree:** Steps 8–10 — detached worktree off the audited commit (`bkit:anchor` `baseSha`), removed before the report finishes; local integration branch ahead of remote ref → landing push carries those commits, so batch builds on them. Fallback: none; do not build the batch in the main checkout.
102
103
  - [ ] **Web:** Steps 13–15 — latest CLI releases, anchor standards, end-of-life dates, vendor support policy. Fallback: versions and days-left recorded as unread.
103
104
 
104
105
  #### Conditional
@@ -106,8 +107,8 @@ Always verify.
106
107
  Verify only when applicable.
107
108
 
108
109
  - [ ] **Bot Dashboard:** Bot publishes a dashboard → Step 3 — blocked and will-not-rebase entries. Fallback: open PRs only; stopped-rebase set `Not Checked`.
109
- - [ ] **Config Validator:** Bot ships one → Step 3 — e.g. `renovate-config-validator`, on the config as the default branch has it. Fallback: read the files; validity `Not Checked`.
110
- - [ ] **Vulnerability Scanner:** Lockfile, image, or requirements file → Step 13 — e.g. Trivy (`trivy fs --include-dev-deps`), OSV-Scanner, `npm audit`, host advisory API. Fallback: one source only; the missing source `Not Checked`.
110
+ - [ ] **Config Validator:** Bot ships one → Step 3 — e.g. `renovate-config-validator`, on the config as the default branch has it, in the bot's read mode (repository, not global / self-hosted) with runner's global options the config needs; wrong mode or missing runner option → errors bot never hits (e.g. Renovate: bare file argument validates as global config; repository mode = `--no-global`, or no argument in a scratch checkout of the default-branch files, + runner `allowedEnv` as `RENOVATE_ALLOWED_ENV`, else false `allowedEnv` errors). Error left in bot's mode = finding, quoted with command; never waived as validator-only. Fallback: read the files; validity `Not Checked`.
111
+ - [ ] **Vulnerability Scanner:** Lockfile, image, or requirements file → Step 13 — e.g. Trivy (`trivy fs --include-dev-deps`), OSV-Scanner, `npm audit`, host advisory API, vendor repository advisories. Fallback: one source only; the missing source `Not Checked`.
111
112
 
112
113
  ### 🔒 Gate 1: Prerequisites Check
113
114
 
@@ -120,26 +121,26 @@ Verify only when applicable.
120
121
 
121
122
  - [ ] **Step 1 — Previous Run:** Audited commit = this report's `bkit:anchor` `baseSha`. Read newest prior report for this slug under `docs/audits/`; its commit = its `bkit:anchor` `baseSha`, else the HEAD its `Previous Run` recorded. Count findings fixed / still open; age of oldest open. Open after 2 runs → re-raise with age. **Done when:** closure rate in `Previous Run`, or run marked baseline.
122
123
  - [ ] **Step 2 — Inventory:** From previous report, list deferred PRs and why. Deferred major whose blocker cleared → first candidate. Select open PRs by head-ref prefix from updater config (`branchPrefix`, default `renovate/`; Dependabot `dependabot/`), never title or author. Token-authored PR on prefix = in scope; record author as evidence, not selector. Per PR: number, title, head ref, head SHA, mergeable, merge state, files, labels, commits behind integration branch, latest CI run (finished only; in-progress ≠ result). **Done when:** `Inventory` names prefix and config source, lists every open PR on prefix, or `none`.
123
- - [ ] **Step 3 — Bot Health:** No dependency updater config = CRITICAL. Examples: Renovate (`.github/renovate.json` or `.github/renovate.json5`; rules example `.bkit/templates/.github/renovate.json5`, self-hosted runner example `.bkit/templates/.github/workflows/bkit-renovate.yml`) or Dependabot (`.github/dependabot.yml`; example `.bkit/templates/.github/dependabot.yml`). See `docs/deployments.md` § Dependency updaters. Both configs = finding; keep one. Bot dashboard: gather-only (bot rewrites). Record blocked entries and will-not-auto-rebase notices → stale until person acts. Bot reads presets from default branch, not integration. Fence / group on integration branch inactive until release carries it. Base-config edits on integration apply only if bot merges base-branch config — option read from default branch. PR proposes what integration config blocks → read default-branch copy before debugging rule; promotion-lag when branches differ = signal. Validate configs as default branch has them. Dashboard + logs: config errors, PRs held by concurrency / hourly limits, awaiting approval, open config-migration PRs. Broken config ≈ quiet week. **Done when:** `Bot Health` holds blocked PRs, config read branch, validator result, held / awaiting-approval items.
124
+ - [ ] **Step 3 — Bot Health:** No dependency updater config = CRITICAL. Examples: Renovate (`.github/renovate.json` or `.github/renovate.json5`; rules example `.bkit/templates/.github/renovate.json5`, self-hosted runner example `.bkit/templates/.github/workflows/bkit-renovate.yml`) or Dependabot (`.github/dependabot.yml`; example `.bkit/templates/.github/dependabot.yml`). See `docs/deployments.md` § Dependency updaters. Both configs = finding; keep one. Bot dashboard: gather-only (bot rewrites). Record blocked entries and will-not-auto-rebase notices → stale until person acts. Bot reads presets from default branch, not integration. Fence / group on integration branch inactive until release carries it. Base-config edits on integration apply only if bot merges base-branch config — option read from default branch. PR proposes what integration config blocks → read default-branch copy before debugging rule; promotion-lag when branches differ = signal. Validate configs as default branch has them, in bot's repository mode (Pre-Flight); error left = finding. Dashboard + logs: config errors, PRs held by concurrency / hourly limits, awaiting approval, open config-migration PRs. Broken config ≈ quiet week. **Done when:** `Bot Health` holds blocked PRs, config read branch, validator result, held / awaiting-approval items.
124
125
  - [ ] **Step 4 — Supply Chain:** Read settings as default branch has them. Record each vs bot best-practice preset and package manager supply-chain page. Bot: minimum release age >= three days (e.g. Renovate `minimumReleaseAge` / `security:minimumReleaseAgeNpm` + `internalChecksFilter: "strict"`; Dependabot `cooldown`); actions / images digest-pinned; abandonment flagging (e.g. Renovate abandonments preset); config-migration PRs enabled, where bot offers them. Package manager, where offered (e.g. pnpm): trust policy no-downgrade, exotic subdeps blocked, explicit install-script allowlist, strict dependency builds; none offered → `N/A` with evidence. Release-age 0 on package manager = recorded trade-off where docs say so; re-deciding = person (`DECISION`). Gap: bot release age does not apply to lockfile-maintenance PRs (package manager makes changes; gate off → no cooldown) → batch with no cooldown; `Cooldown` stands in; `Screen` first. Missing / weaker setting → `Recommendations` row + line to add. **Done when:** `Supply Chain` row per setting: value · standard · verdict.
125
126
  - [ ] **Step 5 — Cooldown:** Package-manager cooldown off or absent → lockfile-maintenance PRs: no automerge; same age wait as other updates before land. Bot opens no lockfile-maintenance PRs (e.g. Dependabot) → those checks `N/A`. Check: lockfile-maintenance block + matching rules → no automerge; each merged lockfile-maintenance since previous run waited >= bot release age (e.g. Renovate `minimumReleaseAge`; newest version publish → merge); open younger → EXCLUDED + eligible date. Automerge groups that rewrite the lockfile: bot release-age check ages named packages only → newest transitive version in the PR's lockfile diff vs bot release age; younger → EXCLUDED + eligible date; automerge with package-manager cooldown off = finding. Same wait on other open non-security PRs: release-age check pending (e.g. `renovate/stability-days`) or newest version younger than bot release age → EXCLUDED + eligible date. Security PRs bypass (Step 13). Lockfile-maintenance automerged or landed before release age = finding. **Done when:** `Cooldown` holds automerge, each merged PR wait, each open PR eligible date.
126
127
  - [ ] **Step 6 — Volume:** Each bot PR = lockfile rewrite + preview + CI; each rebase = another CI. Read on default branch; record what each setting would change; lines to add → project-owned bot config. Grouping: related packages one PR (`group:monorepos`, `group:allNonMajor`, or equivalent). List solo opens that should group + catching rule. Schedule: one window; note outside opens (non-security) + why. Limits: concurrent, hourly, branch-concurrent; record held-back PRs. Rebase policy: conflict-only rebases skip CI on every integration move; unset → record default + count rebase pushes since previous run. Lockfile PR conflicts with most dependency merges; lockfile maintenance own schedule, not every bot run. Automerge: low-risk groups passing CI via platform automerge; lockfile maintenance must not automerge while `Cooldown` off. Majors await dashboard approval; unset → major PR + CI before decision. Scan scope: read bot's detected-file list (e.g. Renovate dashboard) for archived evidence, fixtures, test trees; repo-level ignore list replaces preset defaults (e.g. Renovate `ignorePaths` drops `:ignoreModulesAndTests`) → list paths + PRs touching only them. Setting cutting PRs / CI → `Recommendations` + line + PRs saved this run. **Done when:** `Volume` row per setting: value · what it would change · PRs saved.
127
128
  - [ ] **Step 7 — Classify:** Every PR that is not green. Quote the evidence line from the log. When nothing matches, the class is C.
128
129
  - **A — infrastructure:** preview-deploy wait timeout, registry fetch failure, a cold service refusing, a cancelled runner. Confirm: the same job passed on a neighbouring run or a retry of the same commit.
129
130
  - **B — stale:** outdated lockfile, lockfile config mismatch, a merge conflict, merge state dirty, behind the integration branch. Confirm: behind-count. Being behind says nothing about the dependency itself.
130
- - **C — real:** the failure survives a clean rebase onto the integration branch. Confirm: reproduce it on the batch worktree. Only C keeps a PR out of the batch.
131
+ - **C — real:** the failure survives a clean rebase onto the integration branch. Confirm: reproduce it on the batch worktree before Step 10, calling the test binary directly (e.g. `node_modules/.bin/vitest`); PR bumping the package-manager pin (e.g. `packageManager`) → wrapper exec (e.g. `pnpm exec`) fetches that version and installs, dirtying the worktree. Only C keeps a PR out of the batch.
131
132
  - **C, copy parity:** a test requires two copies of a file to stay identical and the bot updates only the copy it scans (other copy on an ignored path). Fails by design; fix = mirror the change into the other copy. Count recurrence in Step 16.
132
133
 
133
134
  Batch = green + A + B past bot release age. C → EXCLUDED → `Excluded`. Breaking major → DECISION even if IN BATCH. Non-security inside release age → EXCLUDED even if IN BATCH (Step 5). **Done when:** `Classes` row per non-green PR: class · behind-count · quoted evidence · verdict.
134
135
 
135
- - [ ] **Step 8 — Batch:** Detached worktree off integration branch; no new branch. Merge each IN BATCH PR (`git merge --no-edit`); do not push. Lockfile conflict: never text-resolve → integration copy + regenerate (`git checkout <integration> -- <lockfile>`, `git merge --continue` in worktree only). Manifest conflict between batch PRs: higher version wins if not breaking major; else drop lower, record `replaced by #N`. Higher = breaking major → DECISION out; keep lower line, record major not taken. Manifest vs integration (not another PR): higher version line by line. Other conflict → abort merge, drop PR, record why. Record merged commit; install; lockfile-only PRs → refresh exactly named packages. Traps: (1) install rewrites manifests + lockfile (e.g. pnpm catalog, root `package.json`) — uncommitted edits fail frozen CI, pass locally. (2) recursive update rewrites bot-unproposed ranges in big diff — refresh named packages only. (3) partial refresh → split workspace versions; red typecheck / build → clean integration worktree first; fix = one version everywhere incl. shared version catalog (e.g. pnpm catalog). Drift: every manifest byte-identical to merged commit; quote diff command, exit code, output. Do not commit integration branch; never skip-CI marker. Lockfile sources: integrity hash per entry; tarballs from configured registry over HTTPS; both searches empty — quote hits + pulling package. **Done when:** `Batch` holds worktree commit, PRs in/out + why, drift check, both source searches.
136
- - [ ] **Step 9 — Screen:** Batch lockfile diff vs integration — list: new packages (transitive incl.); new install scripts + allowlist prompts; versions published < bot release age before run (lockfile-maintenance first); publisher change or less provenance (trust policy refuses when set); names one edit away from a popular package. Worms via `postinstall` — read every new script. Package still inside release age = Cooldown miss → drop PR, rebuild. Other hits → verdict unchanged; `Recommendations` row + evidence for approver. **Done when:** `Screen` lists each hit, or `none`.
137
- - [ ] **Step 10 — Verify:** `git status --porcelain` empty first (trap (1) in dirty tree). Frozen install → static / unit / integration gate → local E2E per app (store build vs deployed env tests none of batch — use local-build substrate). Beyond green typecheck: framework dependency doctor, where the framework ships one (e.g. `expo-doctor`; SDK-out-of-range packages compile/typecheck then break native/app); single-version check for runtime packages in `Project Instructions` (two versions = trap (3) — report + one version everywhere). Red may be stale install — clear installed deps (e.g. `node_modules`, `.venv`, `target/`), frozen install, re-run before blaming bump. Red check → drop suspect PR, rebuild batch, re-run; no single PR explains → FAIL. **Done when:** `Verification` quotes every command and exit code; `Summary` has PASS, CONCERNS, or FAIL.
136
+ - [ ] **Step 8 — Batch:** Detached worktree off audited commit (`bkit:anchor` `baseSha`), not the remote integration ref — local ahead → landing push carries those commits, so batch builds on them; no new branch. Merge each IN BATCH PR (`git merge --no-edit`); do not push. Lockfile conflict: never text-resolve → integration copy + regenerate (`git checkout <integration> -- <lockfile>`, `git merge --continue` in worktree only). Manifest conflict between batch PRs: higher version wins if not breaking major; else drop lower, record `replaced by #N`. Higher = breaking major → DECISION out; keep lower line, record major not taken. Manifest vs integration (not another PR): higher version line by line. Other conflict → abort merge, drop PR, record why. Record merged commit; install; lockfile-only PRs → refresh exactly named packages. Traps: (1) install rewrites manifests + lockfile (e.g. pnpm catalog, root `package.json`) — uncommitted edits fail frozen CI, pass locally. (2) recursive update rewrites bot-unproposed ranges in big diff — refresh named packages only. (3) partial refresh → split workspace versions; red typecheck / build → clean integration worktree first; fix = one version everywhere incl. shared version catalog (e.g. pnpm catalog). Drift: every manifest byte-identical to merged commit; quote diff command, exit code, output. Do not commit integration branch; never skip-CI marker. Lockfile sources: integrity hash per entry; tarballs from configured registry over HTTPS; both searches empty — quote hits + pulling package. **Done when:** `Batch` holds worktree commit, PRs in/out + why, drift check, both source searches.
137
+ - [ ] **Step 9 — Screen:** Batch lockfile diff vs integration — list: new packages (transitive incl.); new install scripts + allowlist prompts; versions published < bot release age before run (lockfile-maintenance first); publisher change or less provenance (trust policy refuses when set); names one edit away from a popular package; OpenSSF Scorecard score per new package's source repo (e.g. via deps.dev), or `no score` (no linked repo or unscored; not a finding by itself). Worms via `postinstall` — read every new script. Package still inside release age = Cooldown miss → drop PR, rebuild. Other hits → verdict unchanged; `Recommendations` row + evidence for approver. **Done when:** `Screen` lists each hit, or `none`, and a Scorecard score or `no score` per new package.
138
+ - [ ] **Step 10 — Verify:** `git status --porcelain` empty first (trap (1) in dirty tree). Frozen install → static / unit / integration gate → local E2E per app (store build vs deployed env tests none of batch — use local-build substrate). Beyond green typecheck: framework dependency doctor, where the framework ships one (e.g. `expo-doctor`; SDK-out-of-range packages compile/typecheck then break native/app); single-version check for runtime packages in `Project Instructions` (two versions = trap (3) — report + one version everywhere). Red may be stale install — clear installed deps (e.g. `node_modules`, `.venv`, `target/`), frozen install, re-run before blaming bump. Reproduce each red on audited commit without batch (clean worktree off `baseSha`); red there = baseline red → `Verification` + finding, reported apart from batch, not a batch FAIL; batch judged on remaining reds. Remaining red → drop suspect PR, rebuild batch, re-run; no single PR explains → FAIL. **Done when:** `Verification` quotes every command and exit code, and each baseline red; `Summary` has PASS, CONCERNS, or FAIL.
138
139
  - [ ] **Step 11 — Plan:** Report asks and stops — no push, merge, close, label. Direct-push integration branch → batch lands without feature-branch PR. Deployed-boundary checks on push only → landing push = batch meets external services. Never merge bot PRs one-by-one: N sequential → N integration runs + rebase per open PR after each; N one push → one integration run. `Plan` table columns: `Wave`, `PRs`, `Commits`, `Why together, or why apart`, `Pushes`, `CI runs`. Rules: (1) Default one wave — all IN BATCH in one push; commit series (not squash): non-major groups, then one commit per major each with install matching manifests; each commit revert/bisect point; CI once on push head — `Verify` head; earlier commits → frozen install + typecheck. (2) Second wave only for named reason — bisectability not enough; reason = production watch before next land (SDK / framework / native major, store build/submission, DB-client major); extra wave = +push, +integration run, rebase on open PRs — cost in row. (3) Order: security → non-majors → majors; within majors dependency before dependents (language before linter; UI lib + types together). (4) Unmerged automerge group → batch; self-merge first = separate push + rebases — fold in and close at push; merged pre-landing → rebase batch, drop from plan. (5) Close carried PRs right after push with `Excluded` comment → landing commit; bot auto-close next run only — until then may rebase + CI. (6) Land after bot weekly window, no mid-landing opens/rebases, no queued integration push. (7) List excluded/deferred PRs landing will rebase + CI cost; multi-cycle exclude → rebase each later push — recommend close + dashboard when wanted. (8) No skip-CI on dependency change; required checks test batch. Then ordered landing commands: push, close PRs + comments, labels for open. Cost one-by-one vs batch + CI already spent on branch PRs. Post-land watch: error-tracker release, error-rate / crash-free revert thresholds commit-by-commit — do not invent thresholds. **Done when:** `Plan` holds table, commands, both costs, watch.
139
140
  - [ ] **Step 12 — Excluded:** Report lists only — no label, comment, close. C-class + blocked stay open. Report supplies comment text (break + fix) and labels (deferred; major label on major). Comments start with marker in `Project Instructions`; next step finds marker and edits existing comment. Batch-carried PRs → close after land with batch pointer. Breaking major / migration = DECISION → `Recommendations`; real work → issue to open. DECISION evidence (not bare PR): migration guide, breaking changes vs codebase + hit counts, codemods, peer-dep matrix, rollback path. **Done when:** `Excluded` row per out-of-batch PR: verdict · labels · comment text or DECISION evidence.
140
- - [ ] **Step 13 — Unseen:** Bot-blind surfaces. (a) New advisories vs committed baseline — delta script: not in baseline + stale entries; batch fix → remove baseline entry same commit. Security PRs bypass release-age — confirm landing not queued; same as `bkit-audit-security`, do not duplicate. (b) Second vuln source — scanner on lockfile, language requirements, shipped images + host alerts; lockfile scanners reading first doc only → empty without fail — confirm full read. Classify each advisory shipped, server-only, or dev-only, and say whether the vulnerable function is reachable. Destination for an advisory: `bkit-audit-security`. (c) Fix latency — shipped deps: publication → integration land; CRITICAL ≤7d, HIGH ≤30d; breach = finding, list first. (d) Container installing from bot-bumped manifest covered; second install line bypassing manifest = gap, back. (e) Unused deps → `bkit-audit-dead-code`; unmaintained (no release >1y or archived) on new deps since last pass → `Recommendations`. (f) Deprecated / abandoned whole tree — dashboard abandoned + frozen-install warnings. (g) Licences — new deps since last pass; copyleft (GPL, AGPL) or missing in shipped → `Recommendations`. Production tree licence diff vs previous (BSL, SSPL, commercial upgrades too). Shipped app notices regenerated on dep change; bundled fonts/icons/images — licence + ship terms. (h) Off-manifest CLIs — bot reads manifests only. Sources: CI `run:` (global installs, `npx` / `pnpm dlx`, `curl | sh`, system packages), setup-action version inputs (e.g. `supabase/setup-cli` `version:`), build-service CLI pin (e.g. `eas.json` `cli.version`), Dockerfiles, build images, version files (e.g. `.tool-versions`), repo scripts, global-install setup docs. Per CLI: location · pin or `floating` (unpinned, `latest`, open range) · latest release, read live · majors behind · bot-watched (manifest entry or bot rule, e.g. Renovate `customManagers`) yes / no. Floating in CI / build image = finding (unreviewed version each run, bypasses release age). Major behind = finding; end of life → Step 14. Unwatched → `Recommendations`: move into manifest or add bot rule. **Done when:** `Unseen` holds advisories, latency, container install path, deprecated packages, licence diffs, row per off-manifest CLI.
141
+ - [ ] **Step 13 — Unseen:** Bot-blind surfaces. (a) New advisories vs committed baseline — delta script: not in baseline + stale entries; batch fix → remove baseline entry same commit. Security PRs bypass release-age — confirm landing not queued; same as `bkit-audit-security`, do not duplicate. (b) Second vuln source — scanner on lockfile, language requirements, shipped images + host alerts + vendor repository advisories per shipped framework (e.g. `gh api repos/<owner>/<repo>/security-advisories`). Global databases (e.g. `gh api advisories/<id>`, OSV) can lag vendor advisories by days → vendor advisory missing from global database = finding (scanners and CI gate blind to it); already there → count once, no lag finding; no vendor advisories or host API → global only, that source `Not Checked` + reason. Lockfile scanners reading first doc only → empty without fail — confirm full read. Classify each advisory shipped, server-only, or dev-only, and say whether the vulnerable function is reachable. Destination for an advisory: `bkit-audit-security`. (c) Fix latency — shipped deps: publication → integration land; CRITICAL ≤7d, HIGH ≤30d; breach = finding, list first. Per open `CRITICAL` / `HIGH`: earliest bot open date = fixed-version publish + bot release age (none for a security update, per (a)) + next bot run, vs deadline; bot date past deadline = finding before the breach; bot blind to advisory (vendor-only) → no date, advisory itself = finding. (d) Container installing from bot-bumped manifest covered; second install line bypassing manifest = gap, back. (e) Unused deps → `bkit-audit-dead-code`; unmaintained (no release >1y or archived) on new deps since last pass → `Recommendations`. (f) Deprecated / abandoned whole tree — dashboard abandoned + frozen-install warnings. (g) Licences — new deps since last pass; copyleft (GPL, AGPL) or missing in shipped → `Recommendations`. Production tree licence diff vs previous (BSL, SSPL, commercial upgrades too). Shipped app notices regenerated on dep change; bundled fonts/icons/images — licence + ship terms. (h) Off-manifest CLIs — bot reads manifests only. Sources: CI `run:` (global installs, `npx` / `pnpm dlx`, `curl | sh`, system packages), setup-action version inputs (e.g. `supabase/setup-cli` `version:`), build-service CLI pin (e.g. `eas.json` `cli.version`), Dockerfiles, build images, version files (e.g. `.tool-versions`), repo scripts, global-install setup docs. Per CLI: location · pin or `floating` (unpinned, `latest`, open range) · latest release, read live · majors behind · bot-watched (manifest entry or bot rule, e.g. Renovate `customManagers`) yes / no. Floating in CI / build image = finding (unreviewed version each run, bypasses release age). Major behind = finding; end of life → Step 14. Unwatched → `Recommendations`: move into manifest or add bot rule. **Done when:** `Unseen` holds advisories, vendor-advisory gaps, latency + earliest bot dates, container install path, deprecated packages, licence diffs, row per off-manifest CLI.
141
142
  - [ ] **Step 14 — End of Life:** Compare versions in use with the end-of-life catalogue and each vendor's support policy. Record days left for the runtime, the framework, the native SDK, the package manager, the language, the database major, and container base images the project ships (sources in `Project Instructions`). An end of life inside 90 days is DECISION, and the report names the issue to open. **Done when:** `End of Life` row per runtime: version · days left · verdict.
142
- - [ ] **Step 15 — Standards:** Confirm current version per anchor standard; read changes since previous report's version. `Standards Coverage`: one row per bot preset, package-manager supply-chain setting, and Concise Guide rule → checked (name step) / not applicable (why) / finding. Unapplied recommended practice → adopt (propose step) / reject (why) / not applicable. **Done when:** no empty row.
143
+ - [ ] **Step 15 — Standards:** Confirm current version per anchor standard; read changes since previous report's version. `Standards Coverage`: one row per bot preset, package-manager supply-chain setting, and Concise Guide rule → checked (name step) / not applicable (why) / finding. Per bkit anchor (doc section, rule, or issue shape): one row → followed (name step or evidence) / not applicable (why) / finding; departure with no recorded reason (e.g. `docs/decisions.jsonl` row, accepted-divergence entry) = finding. Unapplied recommended practice → adopt (propose step) / reject (why) / not applicable. **Done when:** no empty row.
143
144
  - [ ] **Step 16 — Anything Else:** (a) Since previous report commit — dependency-touching changes (manifests, workspace file, bot configs, workflow `uses:`, Dockerfiles, requirements, runtime-version files, build-image, hosting runtime) → one-line verdict each. (b) ~30 min bounded pass — senior supply-chain gaps no step covers; since previous run: incidents, new `docs/failures.jsonl`, `bkit-audit-logs` reports, vendor status — what no step catches? **Done when:** `Beyond the Steps` holds both passes, or `none`. Class seen on 2 runs → propose step in `Process Improvements`.
144
145
 
145
146
  ### 🔒 Gate 2: Probes Check
@@ -17,7 +17,7 @@
17
17
  > - Any tools mentioned in `Universal Instructions` are examples only; swap for actual tools in final output; drop inapplicable.
18
18
  > - Use dense token-efficient phrasing; no filler / hedging / restating; tables > prose, bullets > paragraphs.
19
19
  > - Read `.agents/docs.md` → load docs for workflow / step / task; skip missing; cite docs inline where required.
20
- > - Every claim cites evidence (`file:line`, or command / query and its output) and tool behind it.
20
+ > - Every claim cites this run's evidence (`file:line`, or command / query and its output) and tool behind it; a previous report's figure, list, or status is a lead, not evidence.
21
21
  > - Read this document once at session start; do not trust session memory; do not re-read on every loop iteration.
22
22
  > - NEVER paste a secret or credential value into the report.
23
23
  > - Historical records (`docs/plans/**`, `docs/archive/**`, `docs/completed.md`) and existing ledger rows are citers only; never edit them.
@@ -17,7 +17,7 @@
17
17
  > - Any tools mentioned in `Universal Instructions` are examples only; swap for actual tools in final output; drop inapplicable.
18
18
  > - Use dense token-efficient phrasing; no filler / hedging / restating; tables > prose, bullets > paragraphs.
19
19
  > - Read `.agents/docs.md` → load docs for workflow / step / task; skip missing; cite docs inline where required.
20
- > - Every claim cites evidence (`file:line`, or command / query and its output) and tool behind it.
20
+ > - Every claim cites this run's evidence (`file:line`, or command / query and its output) and tool behind it; a previous report's figure, list, or status is a lead, not evidence.
21
21
  > - Read this document once at session start; do not trust session memory; do not re-read on every loop iteration.
22
22
  > - Generated registers / graphs → verify-only mode or throwaway copy; never regenerate a tracked file.
23
23
 
@@ -146,7 +146,7 @@ Verify only when applicable.
146
146
  - [ ] **Steps complete:** every step `[x]`, or explicit skip with evidence the surface is absent.
147
147
  - [ ] **Evidence:** every finding cites re-checkable evidence and the tool behind it.
148
148
  - [ ] **Sites read:** every `DUPLICATE` cites `file:line` from an opened file, not a match.
149
- - [ ] **Tree clean:** `git status` clean; tracked registers / graphs unchanged.
149
+ - [ ] **Tree clean:** no tracked file changed by this run (other writers' paths named); tracked registers / graphs unchanged.
150
150
  - [ ] **Gate 2 - Verified:** Every checkbox above independently verified and checked off.
151
151
 
152
152
  ## Findings
@@ -17,7 +17,7 @@
17
17
  > - Any tools mentioned in `Universal Instructions` are examples only; swap for actual tools in final output; drop inapplicable.
18
18
  > - Use dense token-efficient phrasing; no filler / hedging / restating; tables > prose, bullets > paragraphs.
19
19
  > - Read `.agents/docs.md` → load docs for workflow / step / task; skip missing; cite docs inline where required.
20
- > - Every claim cites evidence (`file:line`, or command / query and its output) and tool behind it.
20
+ > - Every claim cites this run's evidence (`file:line`, or command / query and its output) and tool behind it; a previous report's figure, list, or status is a lead, not evidence.
21
21
  > - Read this document once at session start; do not trust session memory; do not re-read on every loop iteration.
22
22
  > - NEVER append or edit a ledger line. Close, regression, triage, and knowledge rows live in `Expected` only. `ts` is minted at append (`.bkit/jsonl.md` § Write Idiom), never written in this report.
23
23
  > - A security row (`category: "security"`, or an auth, secrets, or data-exposure hole) carries key, verdict, and severity only. `file:line` stays in the ledger row. A public repo uses a draft repository security advisory, not this report.
@@ -17,7 +17,7 @@
17
17
  > - Any tools mentioned in `Universal Instructions` are examples only; swap for actual tools in final output; drop inapplicable.
18
18
  > - Use dense token-efficient phrasing; no filler / hedging / restating; tables > prose, bullets > paragraphs.
19
19
  > - Read `.agents/docs.md` → load docs for workflow / step / task; skip missing; cite docs inline where required.
20
- > - Every claim cites evidence (`file:line`, or command / query and its output) and tool behind it.
20
+ > - Every claim cites this run's evidence (`file:line`, or command / query and its output) and tool behind it; a previous report's figure, list, or status is a lead, not evidence.
21
21
  > - Read this document once at session start; do not trust session memory; do not re-read on every loop iteration.
22
22
  > - never `.skip`, quarantine, delete an assertion, or widen a timeout to green a lane. A timeout named before the trace is read is not a verdict.
23
23
  > - Repro and repeat runs use the suite's test tenant only; NEVER production tenants, payments, or mail.
@@ -42,6 +42,7 @@
42
42
  - Order-dependence and root-cause location (e.g. iDFlakies, Lam et al., ICST 2019; Ziftci and Cavalcanti, "De-Flake Your Tests", ICSME 2020).
43
43
  - Browser-runner retries, flaky status, repeat-each, fail-on-flaky (e.g. Playwright); record the version.
44
44
  - Unit-runner shuffle, seed, and retry (e.g. Vitest `sequence.shuffle` / `sequence.seed` / `retry`, pytest-randomly); record the version.
45
+ - bkit block of `docs/testing.md` (bkit version from `.bkit/version`): § Which resources may each lane touch?; § E2E on PRs: run the full suite, watch which DB it hits; § Concurrent invocations collide on disk, not just in the database
45
46
 
46
47
  ### Scope
47
48
 
@@ -120,7 +121,7 @@ Verify only when applicable.
120
121
  - [ ] **Step 10 — Mechanism:** Teardown deletes rows event points at → most diagnoses corroborated by mechanism, not row. Distinguish "fits" vs "proven". One root-cause class: async wait, concurrency, order dependency, shared fixture state, resource leak, network, external service, time and date, randomness, unordered collection, infrastructure, or other (describe). Luo et al.: async wait, concurrency, order dependency ≈ 78% classified. Per-class counts across open flakes = series. Before timing mechanism → floating-promise/missing-await rules on spec/helpers (e.g. `@typescript-eslint/no-floating-promises`, `playwright/missing-playwright-await`); record hits so missing `await` is not hidden behind timing. Real defect behind flake → verdict `BUG`. **Done when:** class, what is not established, await hits in `Mechanism`.
121
122
  - [ ] **Step 11 — Fix:** Name cause and change that removes it; making change = next step. Never `.skip`, quarantine, delete assertion, widen timeout. Before trusting the repaired test, the next step puts the bug back, watches the test go red, restores, and confirms `git diff` on the source file is empty. That is not optional on anything touching row-level access control or a cross-tenant boundary. Shared helper/fixture change = signature change: references first. One-spec fix via shared fixture → next unattributable failure. Put bug back proves deterministic fix. Flaky fix also repeats Step 4: at rate p, ~3/p runs with no failure → 95% confidence rate below p (rule of three); report states count. One verdict from `Verdicts`. **Done when:** verdict, change, or why none, run count in `Fix`.
122
123
  - [ ] **Step 12 — Route:** No finding without re-checkable evidence: `file:line`, or query/run id/measurement + tool. Language-server empty inconclusive → text search corroboration. Report names destinations (sending = next step); the audit agent writes `docs/failures.jsonl` rows. Every flake → `docs/failures.jsonl` row (class, rate, age). Named mechanism → recommended fix + proof runs (Step 11). Real defect → own issue + ledger row. Fix waits on decision → named decision. Mechanism error tracker also saw → `bkit-audit-logs`. Retry-pass/warning CI annotated → `bkit-audit-annotations`. Missing trace/capture, third-party stub, missing `await` → recommended change (Steps 5, 10). **Done when:** every finding's Expected names destination, including ledger row.
123
- - [ ] **Step 13 — Standards:** Confirm current version per anchor standard; read changes since previous report's version. `Standards Coverage`: row per Luo et al. class + per flake feature the runner offers (e.g. retries, flaky status, repeat-each, fail-on-flaky, sequence shuffle) → checked (name step) / not applicable (why) / finding. Unapplied recommended practice → adopt (propose step) / reject (why) / not applicable. Do not change steps during run. **Done when:** no empty row.
124
+ - [ ] **Step 13 — Standards:** Confirm current version per anchor standard; read changes since previous report's version. `Standards Coverage`: row per Luo et al. class + per flake feature the runner offers (e.g. retries, flaky status, repeat-each, fail-on-flaky, sequence shuffle) → checked (name step) / not applicable (why) / finding. Per bkit anchor (doc section, rule, or issue shape): one row → followed (name step or evidence) / not applicable (why) / finding; departure with no recorded reason (e.g. `docs/decisions.jsonl` row, accepted-divergence entry) = finding. Unapplied recommended practice → adopt (propose step) / reject (why) / not applicable. Do not change steps during run. **Done when:** no empty row.
124
125
  - [ ] **Step 14 — Anything Else:** (a) Since previous report's commit, everything touching test stability (browser/unit config: retries, workers, timeouts, trace; shared fixtures/helpers; CI workflows/runners; preview/DB for E2E; device flows) → one-line verdict each. (b) Bounded ~30 min: senior test-infra flags no step asks; read since previous run (incidents, `bkit-audit-logs`, new `docs/failures.jsonl`, vendor status) — what would no step catch? **Done when:** `Beyond the Steps` holds both passes, or `none`. Class seen on 2 runs → propose step in `Process Improvements`.
125
126
 
126
127
  ### 🔒 Gate 2: Probes Check
@@ -17,7 +17,7 @@
17
17
  > - Any tools mentioned in `Universal Instructions` are examples only; swap for actual tools in final output; drop inapplicable.
18
18
  > - Use dense token-efficient phrasing; no filler / hedging / restating; tables > prose, bullets > paragraphs.
19
19
  > - Read `.agents/docs.md` → load docs for workflow / step / task; skip missing; cite docs inline where required.
20
- > - Every claim cites evidence (`file:line`, or command / query and its output) and tool behind it.
20
+ > - Every claim cites this run's evidence (`file:line`, or command / query and its output) and tool behind it; a previous report's figure, list, or status is a lead, not evidence.
21
21
  > - Read this document once at session start; do not trust session memory; do not re-read on every loop iteration.
22
22
  > - never build, wire, or land a gate; never edit workflows, hooks, allowlists, or gate scripts.
23
23
  > - NEVER weaken an existing gate — no skip, rule downgrade, lowered floor, or broader ignore — so a candidate looks adoptable.
@@ -42,6 +42,9 @@ Maintenance sweeps find breakage; none prevent the next. Sibling audits repeat
42
42
  - Required status checks and a merge queue (e.g. GitHub rulesets).
43
43
  - OpenSSF Scorecard checks, at the Scorecard version read.
44
44
  - DORA continuous integration capability.
45
+ - bkit block of `docs/testing.md` (bkit version from `.bkit/version`): § CI test tiers (the bkit scaffold); § Accepted divergences; § Step names say whether a red is real; § Composite actions: two rules that fail silently
46
+ - bkit block of `docs/deployments.md` (bkit version from `.bkit/version`): § Commit-message force tags; § Workflow-file limits GitHub does not enforce visibly; § Ship gate + bookkeeping
47
+ - bkit block of `docs/design.md` (bkit version from `.bkit/version`): § Machine-checked reference
45
48
 
46
49
  ### Scope
47
50
 
@@ -137,7 +140,7 @@ Verify only when applicable.
137
140
  - [ ] **Step 11 — Re-Review:** Gate built and reverted is not settled. Re-read each rejection in `docs/decisions.jsonl` and prior reports; check whether reasoning still holds. Two decisions on one guard, neither retired → criterion cannot be judged = finding. Reasoning no longer holds → score again into `Scores`, with `docs/decisions.jsonl` row that would retire old decision. **Done when:** `Rejections` row per reverted or rejected gate: still holds?
138
141
  - [ ] **Step 12 — Health:** Per landed gate, since previous report: fire rate, true-positive rate, overrides/bypasses, own flakes, runtime. Never fired = perfectly preventive or hollow — indistinguishable outside; fresh positive control (`Positive Control`) tells apart. Never fired or hollow → name that control as next step. Every `bkit:gate:*` script in `package.json` = landed gate; workflows, agents, pre-push run only these → expand each (follow every nested `pnpm <script>` to its body) → list each command. Hollow → `CRITICAL`: script missing, empty, or runs nothing that can fail (e.g. `echo`, `true`, `|| exit 0`); command exits 0 when its tool absent (e.g. `command -v … || echo skipping`) = that check hollow → name tool and install step. **Done when:** `Health` row per landed gate, or none landed; each `bkit:gate:*` row lists every command and which can skip.
139
142
  - [ ] **Step 13 — Allowlists:** Every allowlist entry (known-offender maps, ratchet baseline, dead-code ignores) past date or with cited issue closed. Each lapsed = finding: renew with reason, or fix offender. Per ratchet baseline, compare entry count vs previous report: grew, or entry with no owner = finding. **Done when:** `Allowlists` lists every lapsed, grown, or ownerless entry, or `none`.
140
- - [ ] **Step 14 — Standards:** Confirm current version per anchor standard; read changes since previous report's version. `Standards Coverage`: one row per anchor practice (fitness functions, static-analysis false-positive bar, new-code gating, ratchets, policy as code, rulesets and merge queue, DORA CI capability) and per OpenSSF Scorecard check → checked (name step) / not applicable (why) / finding. Unapplied recommended practice → adopt (propose step) / reject (why) / not applicable. **Done when:** no empty row.
143
+ - [ ] **Step 14 — Standards:** Confirm current version per anchor standard; read changes since previous report's version. `Standards Coverage`: one row per anchor practice (fitness functions, static-analysis false-positive bar, new-code gating, ratchets, policy as code, rulesets and merge queue, DORA CI capability) and per OpenSSF Scorecard check → checked (name step) / not applicable (why) / finding. Per bkit anchor (doc section, rule, or issue shape): one row → followed (name step or evidence) / not applicable (why) / finding; departure with no recorded reason (e.g. `docs/decisions.jsonl` row, accepted-divergence entry) = finding. Unapplied recommended practice → adopt (propose step) / reject (why) / not applicable. **Done when:** no empty row.
141
144
  - [ ] **Step 15 — Anything Else:** (a) Everything changed since previous report's commit touching gating (guards and contract tests, hook directory, gate scripts, workflow jobs, rulesets and branch protection, allowlists) → one-line verdict each. (b) Bounded ~30 min pass: what senior build and developer-productivity engineer flags that no step asks; read since previous run (incidents, `bkit-audit-logs` reports, new `docs/failures.jsonl` rows, vendor status incidents) — what would no step catch? **Done when:** `Beyond the Steps` holds both passes, or `none`. Class seen on 2 runs → propose step in `Process Improvements`.
142
145
 
143
146
  ### 🔒 Gate 2: Probes Check