Refine self-improvement review guidance

Narrow IsHandled, label-scope, UI-handler, checkpoint, and bulk-operation guidance to evidence-backed false-positive boundaries.

Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>

microsoft/skills/review/al-events-review.md

@@ -51,7 +51,7 @@ When the post-conflict worklist is empty because no applicable events knowledge
 
 The following targeted checks map diff signals to specific `events` articles. Treat each as a candidate-selection cue: when the signal appears in the changed code, add the named article to the worklist and evaluate it in Action.
 
-- `IsHandled` raised without an immediately preceding `IsHandled := false;`, or one `IsHandled` variable reused across several raises with no reset between them — `initialize-ishandled-to-false-before-publishing`.
+- An `IsHandled` value that can carry over as `true` (reused after an earlier raise, input/global/field, or otherwise seeded) is passed to another publisher without a reset — `initialize-ishandled-to-false-before-publishing`. Do not match a single raise using a fresh local Boolean, or a later raise reached only after `if IsHandled then exit;`.
 - `if IsHandled then exit;` in a routine that also raises a paired `OnAfter…` event later, so the after-event is skipped whenever the call is handled — `preserve-onafter-execution-when-ishandled-skips-the-body`.
 - Any parameter added to a public Business/Integration event procedure, regardless of position; do not flag additions or reordering on `local`/`internal` publishers merely because a new parameter was not appended — `add-new-event-parameters-at-the-end`.
 - A shipped Business/Integration event renamed or removed, or an existing parameter renamed, removed, retyped, or changed to/from `var`, based on the mistaken assumption that `local` or `internal` prevents dependent subscription; parameter order alone is not a subscriber-contract violation — `treat-local-and-internal-events-as-subscriber-contracts`.

microsoft/skills/review/al-performance-review.md

@@ -47,7 +47,8 @@ Apply these targeted cues even when simple token overlap would rank the article
 
 - Worklist `use-setautocalcfields-for-per-row-flowfields.md` when a record loop calls `CalcFields`, or when every row reads the same FlowField for a comparison, branch, or per-record action. Worklist `calcsums-instead-of-calcfields-in-loop.md` instead when the loop only accumulates one set total.
 - Worklist `hidden-flowfields-still-calculate-before-bc26-opt-in.md` when a page control directly sources a FlowField and sets `Visible = false` or a visibility expression. Suppress it when the target is known to have BC26's **Calculate only visible FlowFields** feature enabled, or when the FlowField is cheap and intentionally preloaded.
-- Worklist `avoid-commit-inside-loops.md` only when `Commit()` is inside a record-iteration body or a helper invoked once per row. Do not match one `Commit()` after a bounded checkpoint helper returns, a `Commit()` outside iteration, or comments and documentation that merely mention commits.
+- Worklist `avoid-commit-inside-loops.md` when `Commit()` is inside a record-iteration body or a checkpoint loop lacks persisted progress that excludes completed work on retry. Do not match a commit after a complete business unit when the same transaction persists a restart-safe watermark/state and errors propagate. Still match a full-tail `FindSet` with periodic commits as unbounded retrieval; restart safety does not make it `TOP X`.
+- Worklist `prefer-modifyall-over-per-row-modify.md` for a constant-assignment `Modify(false)` loop with no validation or per-row semantics. Worklist `triggers-and-media-field-regress-modifyall.md` when table trigger code, related subscribers, security filtering, `Media`/`MediaSet`, or companion fields affect a bulk path. A progress dialog does not generically exempt a loop; accept it only when the equivalent bulk call already falls back to individual operations and semantics are preserved.
 - Worklist `avoid-cloning-records-before-modify-delete-in-loops.md` when an iteration calls `Copy` or `RecordRef.GetTable` before `Modify`/`Delete`, or passes the iterated record without `var` to a helper that writes that record. Do not worklist it from `Modify`, `Delete`, or `RecordRef` alone; exclude a direct write on the iterator, a read-only copy, a temporary record, a different target table, and a `RecordRef` opened and iterated directly.
 - Worklist `use-tryfunction-for-error-catching-not-rollback.md` only when writes occur inside a try method and the code or surrounding flow expects an error to roll them back. A bare try-method call whose Boolean result is ignored belongs exclusively to `error-handling/ignored-tryfunction-return-disables-try-semantics.md`; do not worklist the performance article from that call shape alone.
 - For `LockTable` in a pure read helper, select exactly one owner. Use `do-not-locktable-in-read-only-procedure.md` when the helper needs no stronger isolation and should remove the lock. Use `prefer-readisolation-over-locktable-for-reads.md` instead when the code explicitly requires committed-read semantics and `ReadIsolation` is the replacement. Never emit both findings for the same call.
@@ -74,7 +75,7 @@ Set `confidence` to:
 
 After evaluating each worklist entry, also consider whether the diff exhibits a performance defect the agent recognises from its general AL knowledge that no knowledge file in the worklist covers. Such candidates are agent findings within this skill's domain — emit them with `references: []`, an `id` slug prefixed with `agent:`, `confidence` capped at `medium`, `severity` capped at `minor` (agent findings are advisory and non-gating), and a `message` that is self-contained (describing both the issue and a concrete recommendation, since there is no knowledge-file footer for the consumer to fall back on). Hold every candidate to the precision bar in `skills/do.md` (*Agent findings*): emit only a concrete, material performance defect a knowledgeable BC reviewer would agree is wrong — steelman it first and drop anything stylistic, speculative, dependent on code outside the diff, or merely a valid alternative; when in doubt, omit. The scope is strictly performance; defects outside this domain belong to other leaves and MUST NOT be emitted here. Before emitting, check the worklist for a knowledge file that matches the candidate — if one exists, upgrade the candidate to a knowledge-backed finding instead. See `skills/do.md` for the full contract.
 
-For every emitted finding, decide whether the fix is mechanical. A fix is mechanical when it is small, local, and unambiguous from the diff context (for example: delete unreachable lines; replace `Count() > 0` with `not IsEmpty()`; move a local `Label` to object scope; add a missing `ToolTip`, `OptionCaption`, or `DataClassification`; replace a string-concatenated `Error` with a Label-backed call; change an over-broad permission token; or add an obvious `else`/guard branch). For mechanical findings, emit `findings[].suggested-code` with the literal replacement for the source lines indicated by `location`. The payload must be a verbatim replacement — no diff markers, no fences, no commentary — that the consumer can render as a one-click suggestion. When a `.good.al` companion exists and the diff context matches the `.bad.al` shape, adapt the `.good.al` replacement into `suggested-code`.
+For every emitted finding, decide whether the fix is mechanical. A fix is mechanical when it is small, local, and unambiguous from the diff context (for example: delete unreachable lines; replace `Count() > 0` with `not IsEmpty()`; add a missing `ToolTip`, `OptionCaption`, or `DataClassification`; replace a string-concatenated `Error` with a Label-backed call; change an over-broad permission token; or add an obvious `else`/guard branch). For mechanical findings, emit `findings[].suggested-code` with the literal replacement for the source lines indicated by `location`. The payload must be a verbatim replacement — no diff markers, no fences, no commentary — that the consumer can render as a one-click suggestion. When a `.good.al` companion exists and the diff context matches the `.bad.al` shape, adapt the `.good.al` replacement into `suggested-code`.
 
 Omit `suggested-code` only when the appropriate fix depends on context the skill cannot determine, when multiple defensible replacements exist, or when the fix spans non-contiguous code. If a finding is mechanical-looking but you omit `suggested-code`, set `findings[].suggested-code-omission-reason` to a short explanation. See `skills/do.md` for the full contract.
 

microsoft/skills/review/al-privacy-review.md

@@ -70,7 +70,7 @@ Set `confidence` to:
 
 This leaf emits only knowledge-backed privacy findings. Do NOT emit reference-less `agent:` findings in this domain: online evaluation shows the privacy agent-finding channel yields almost no accepted findings and a high volume of dismissed noise, so a privacy concern that no worklist knowledge file covers is omitted here rather than emitted with `references: []`. When you spot a material privacy defect no article covers, the durable fix is to add a knowledge article in BCQuality (per the online-eval self-improvement loop) so this leaf can cite it — not a one-off reference-less finding. Before treating a candidate as uncovered, check the worklist for a knowledge file that matches it; if one exists, emit it as a knowledge-backed finding. See `skills/do.md` for the full contract.
 
-For every emitted finding, decide whether the fix is mechanical. A fix is mechanical when it is small, local, and unambiguous from the diff context (for example: delete unreachable lines; replace `Count() > 0` with `not IsEmpty()`; move a local `Label` to object scope; add a missing `ToolTip`, `OptionCaption`, or `DataClassification`; replace a string-concatenated `Error` with a Label-backed call; change an over-broad permission token; or add an obvious `else`/guard branch). For mechanical findings, emit `findings[].suggested-code` with the literal replacement for the source lines indicated by `location`. The payload must be a verbatim replacement — no diff markers, no fences, no commentary — that the consumer can render as a one-click suggestion. When a `.good.al` companion exists and the diff context matches the `.bad.al` shape, adapt the `.good.al` replacement into `suggested-code`.
+For every emitted finding, decide whether the fix is mechanical. A fix is mechanical when it is small, local, and unambiguous from the diff context (for example: delete unreachable lines; replace `Count() > 0` with `not IsEmpty()`; add a missing `ToolTip`, `OptionCaption`, or `DataClassification`; replace a string-concatenated `Error` with a Label-backed call; change an over-broad permission token; or add an obvious `else`/guard branch). For mechanical findings, emit `findings[].suggested-code` with the literal replacement for the source lines indicated by `location`. The payload must be a verbatim replacement — no diff markers, no fences, no commentary — that the consumer can render as a one-click suggestion. When a `.good.al` companion exists and the diff context matches the `.bad.al` shape, adapt the `.good.al` replacement into `suggested-code`.
 
 Omit `suggested-code` only when the appropriate fix depends on context the skill cannot determine, when multiple defensible replacements exist, or when the fix spans non-contiguous code. If a finding is mechanical-looking but you omit `suggested-code`, set `findings[].suggested-code-omission-reason` to a short explanation. See `skills/do.md` for the full contract.
 

microsoft/skills/review/al-security-review.md

@@ -70,7 +70,7 @@ Set `confidence` to:
 
 After evaluating each worklist entry, also consider whether the diff exhibits a security defect the agent recognises from its general AL knowledge that no knowledge file in the worklist covers. Such candidates are agent findings within this skill's domain — emit them with `references: []`, an `id` slug prefixed with `agent:`, `confidence` capped at `medium`, `severity` capped at `minor` (agent findings are advisory and non-gating), and a `message` that is self-contained (describing both the issue and a concrete recommendation, since there is no knowledge-file footer for the consumer to fall back on). Hold every candidate to the precision bar in `skills/do.md` (*Agent findings*): emit only a concrete, material security defect a knowledgeable BC reviewer would agree is wrong — steelman it first and drop anything stylistic, speculative, dependent on code outside the diff, or merely a valid alternative; when in doubt, omit. The scope is strictly security; defects outside this domain belong to other leaves and MUST NOT be emitted here. Before emitting, check the worklist for a knowledge file that matches the candidate — if one exists, upgrade the candidate to a knowledge-backed finding instead. See `skills/do.md` for the full contract.
 
-For every emitted finding, decide whether the fix is mechanical. A fix is mechanical when it is small, local, and unambiguous from the diff context (for example: delete unreachable lines; replace `Count() > 0` with `not IsEmpty()`; move a local `Label` to object scope; add a missing `ToolTip`, `OptionCaption`, or `DataClassification`; replace a string-concatenated `Error` with a Label-backed call; change an over-broad permission token; or add an obvious `else`/guard branch). For mechanical findings, emit `findings[].suggested-code` with the literal replacement for the source lines indicated by `location`. The payload must be a verbatim replacement — no diff markers, no fences, no commentary — that the consumer can render as a one-click suggestion. When a `.good.al` companion exists and the diff context matches the `.bad.al` shape, adapt the `.good.al` replacement into `suggested-code`.
+For every emitted finding, decide whether the fix is mechanical. A fix is mechanical when it is small, local, and unambiguous from the diff context (for example: delete unreachable lines; replace `Count() > 0` with `not IsEmpty()`; add a missing `ToolTip`, `OptionCaption`, or `DataClassification`; replace a string-concatenated `Error` with a Label-backed call; change an over-broad permission token; or add an obvious `else`/guard branch). For mechanical findings, emit `findings[].suggested-code` with the literal replacement for the source lines indicated by `location`. The payload must be a verbatim replacement — no diff markers, no fences, no commentary — that the consumer can render as a one-click suggestion. When a `.good.al` companion exists and the diff context matches the `.bad.al` shape, adapt the `.good.al` replacement into `suggested-code`.
 
 Omit `suggested-code` only when the appropriate fix depends on context the skill cannot determine, when multiple defensible replacements exist, or when the fix spans non-contiguous code. If a finding is mechanical-looking but you omit `suggested-code`, set `findings[].suggested-code-omission-reason` to a short explanation. See `skills/do.md` for the full contract.
 

microsoft/skills/review/al-style-review.md

@@ -60,7 +60,7 @@ When the post-conflict worklist is empty because no applicable style knowledge e
 
 For each worklist entry, evaluate the diff against the file's `## Best Practice` and `## Anti Pattern` sections. Style findings rarely reach `blocker` — reserve it for cases where the knowledge file documents a platform-level requirement (for example, API page property constraints the OData runtime rejects). Most style findings are `minor` or `info`; egregious misuse (`Error` with pre-built Text losing translation and telemetry classification) may reach `major`.
 
-Severity calibration — a formal analyzer already flags the mechanical presence/naming conventions (the `this` keyword AA0248, approved label suffixes AA0074, variable-declaration order by type AA0021, a missing `ToolTip`, required parentheses). On those, BCQuality's value is the *explanation* of why the rule exists, not a second gate; emit them at `info` so a consumer that gates on severity does not re-flag what CodeCop/AppSourceCop already reports. Reserve `minor` for style issues with concrete downstream impact the analyzer does not catch — a `Label` declared at procedure-local instead of object scope (no analyzer enforces label scope, and mis-scoped Labels are fragile in the translation pipeline), lost translation or telemetry classification from a string-built `Error`, an `OptionCaption` that does not match its `OptionMembers`, a misleading named invocation. This keeps the domain's default output advisory and prevents analyzer-redundant noise from competing with substantive review.
+Severity calibration — a formal analyzer already flags the mechanical presence/naming conventions (the `this` keyword AA0248, approved label suffixes AA0074, variable-declaration order by type AA0021, a missing `ToolTip`, required parentheses). On those, BCQuality's value is the *explanation* of why the rule exists, not a second gate; emit them at `info` so a consumer that gates on severity does not re-flag what CodeCop/AppSourceCop already reports. Reserve `minor` for style issues with concrete downstream impact the analyzer does not catch — lost translation or telemetry classification from a string-built `Error`, an `OptionCaption` that does not match its `OptionMembers`, or a misleading named invocation. A procedure-local `Label` is valid and is not a correctness or localization finding; an explicit repository preference for object scope is at most low-severity maintainability guidance. This keeps the domain's default output advisory and prevents analyzer-redundant noise from competing with substantive review.
 
 Set `confidence` to:
 
@@ -70,7 +70,7 @@ Set `confidence` to:
 
 After evaluating each worklist entry, also consider whether the diff exhibits a style defect the agent recognises from its general AL knowledge that no knowledge file in the worklist covers. Such candidates are agent findings within this skill's domain — emit them with `references: []`, an `id` slug prefixed with `agent:`, `confidence` capped at `medium`, `severity` capped at `minor` (agent findings are advisory and non-gating), and a `message` that is self-contained (describing both the issue and a concrete recommendation, since there is no knowledge-file footer for the consumer to fall back on). Hold every candidate to the precision bar in `skills/do.md` (*Agent findings*): emit only a clear, widely-accepted AL style violation with a concrete basis a knowledgeable BC reviewer would agree on — steelman it first and drop personal preference, speculation, and any single defensible formatting choice among several; when in doubt, omit. The scope is strictly style — naming, labelling, formatting, and analyzer-adjacent conventions. A correctness, logic, data-integrity, or contract defect is NOT a style finding even when it can be reworded as a convention: a method that mutates a shared `Record`'s filters, an unfiltered `DeleteAll`, a violated interface contract, or a wrong boolean guard are behavioural defects, not conventions — do not emit them here under a style framing. If a specific domain leaf covers the concern (performance, security, error-handling, …) it belongs there; if no knowledge file in any domain covers it, it belongs to the `al-code-review` super-skill's cross-cutting self-review agent channel (`from-sub-skill: "agent"`, `severity` capped at `minor`), not to this leaf. A reliable test: if you cannot cite a style `## Best Practice`/`## Anti Pattern` for the concern, it is very likely not a style finding. Before emitting, check the worklist for a knowledge file that matches the candidate — if one exists, upgrade the candidate to a knowledge-backed finding instead. See `skills/do.md` for the full contract.
 
-For every emitted finding, decide whether the fix is mechanical. A fix is mechanical when it is small, local, and unambiguous from the diff context (for example: delete unreachable lines; replace `Count() > 0` with `not IsEmpty()`; move a local `Label` to object scope; add a missing `ToolTip`, `OptionCaption`, or `DataClassification`; replace a string-concatenated `Error` with a Label-backed call; change an over-broad permission token; or add an obvious `else`/guard branch). For mechanical findings, emit `findings[].suggested-code` with the literal replacement for the source lines indicated by `location`. The payload must be a verbatim replacement — no diff markers, no fences, no commentary — that the consumer can render as a one-click suggestion. When a `.good.al` companion exists and the diff context matches the `.bad.al` shape, adapt the `.good.al` replacement into `suggested-code`.
+For every emitted finding, decide whether the fix is mechanical. A fix is mechanical when it is small, local, and unambiguous from the diff context (for example: delete unreachable lines; replace `Count() > 0` with `not IsEmpty()`; add a missing `ToolTip`, `OptionCaption`, or `DataClassification`; replace a string-concatenated `Error` with a Label-backed call; change an over-broad permission token; or add an obvious `else`/guard branch). For mechanical findings, emit `findings[].suggested-code` with the literal replacement for the source lines indicated by `location`. The payload must be a verbatim replacement — no diff markers, no fences, no commentary — that the consumer can render as a one-click suggestion. When a `.good.al` companion exists and the diff context matches the `.bad.al` shape, adapt the `.good.al` replacement into `suggested-code`.
 
 Omit `suggested-code` only when the appropriate fix depends on context the skill cannot determine, when multiple defensible replacements exist, or when the fix spans non-contiguous code. If a finding is mechanical-looking but you omit `suggested-code`, set `findings[].suggested-code-omission-reason` to a short explanation. See `skills/do.md` for the full contract.
 

microsoft/skills/review/al-testing-review.md

@@ -50,7 +50,7 @@ The following targeted checks cover every current `testing` article. Treat each
 - A permission-sensitive test uses `TestPermissions = Disabled`, claims to test a restricted user without `"Permissions Mock"`/`"Library - Lower Permissions"`, or declares `[TestPermissions(...)]` without applying that context — `permission-tests-must-lower-the-execution-context`.
 - Test fixture code manually calls `Init`/`Insert`, invents keys or prerequisite records, or bypasses available `LibrarySales`, `LibraryPurchase`, `LibraryERM`, `LibraryInventory`, `LibraryRandom`, or equivalent library codeunits — `use-library-codeunits-for-test-fixtures`.
 - `asserterror` is added or changed without a following `Assert.ExpectedError`, `Assert.ExpectedErrorCode`, or a purpose-built assertion such as `ExpectedTestFieldError` — `asserterror-needs-expectederror-and-code`.
-- A test path raises UI, `[HandlerFunctions(...)]` does not exactly match the invoked handlers, a handler hardcodes replies instead of using enqueue/dequeue expectations, or `LibraryVariableStorage.Clear`/`AssertEmpty` is missing — `ui-handlers-in-tests`.
+- A test path raises UI and `[HandlerFunctions(...)]` does not match the invoked handlers, or the test has no meaningful evidence of the UI result (for example, it treats a Boolean set before the action as proof of success) — `ui-handlers-in-tests`. A capture/reset/assert-after-`RunModal` pattern is valid. Enqueue/dequeue and `AssertEmpty` are required only when order, count, text, replies, or a scripted sequence is part of the contract.
 
 Once the candidate worklist is known, resolve layer-precedence conflicts per READ. Drop lower-precedence files whose normative guidance (`## Best Practice` or `## Anti Pattern`) directly contradicts a higher-precedence candidate, and record each dropped file in `suppressed` with `reason: "layer-precedence"`. Files that would have been candidates but are hidden because their layer is disabled in consumer configuration are recorded with `reason: "configuration"`. Files that never became candidates are NOT recorded in `suppressed`.
 
@@ -64,6 +64,8 @@ For each worklist entry, evaluate the diff against the file's `## Best Practice`
 - When the diff contains code that contradicts a Best Practice without being a full anti-pattern, emit `minor` with the same reference shape.
 - Applicability alone is not a finding. Emit `info` only for a concrete, non-actionable observation the article explicitly defines; otherwise emit nothing when no violation is present.
 
+For `ui-handlers-in-tests`, use `major` when missing or incorrectly listed handlers make the test fail at runtime. Use `minor` when the test executes but lacks a meaningful semantic postcondition, including a pre-set Boolean used as proof. Do not escalate solely because a handler does not use queue storage or asserts inside the handler.
+
 Set `confidence` to:
 
 - `high` when the detection is based on an unambiguous pattern match (attribute, handler declaration, assertion sequence, or fixture call).

microsoft/skills/review/al-ui-review.md

@@ -63,7 +63,7 @@ Set `confidence` to:
 
 This leaf emits only knowledge-backed UI and accessibility findings. Do NOT emit reference-less `agent:` findings in this domain: online evaluation shows the UI/accessibility agent-finding channel yields almost no accepted findings and a high volume of dismissed noise, so a UI or accessibility concern that no worklist knowledge file covers is omitted here rather than emitted with `references: []`. When you spot a material UI or accessibility defect no article covers, the durable fix is to add a knowledge article in BCQuality (per the online-eval self-improvement loop) so this leaf can cite it — not a one-off reference-less finding. Before treating a candidate as uncovered, check the worklist for a knowledge file that matches it; if one exists, emit it as a knowledge-backed finding. See `skills/do.md` for the full contract.
 
-For every emitted finding, decide whether the fix is mechanical. A fix is mechanical when it is small, local, and unambiguous from the diff context (for example: delete unreachable lines; replace `Count() > 0` with `not IsEmpty()`; move a local `Label` to object scope; add a missing `ToolTip`, `OptionCaption`, or `DataClassification`; replace a string-concatenated `Error` with a Label-backed call; change an over-broad permission token; or add an obvious `else`/guard branch). For mechanical findings, emit `findings[].suggested-code` with the literal replacement for the source lines indicated by `location`. The payload must be a verbatim replacement — no diff markers, no fences, no commentary — that the consumer can render as a one-click suggestion. When a `.good.al` companion exists and the diff context matches the `.bad.al` shape, adapt the `.good.al` replacement into `suggested-code`.
+For every emitted finding, decide whether the fix is mechanical. A fix is mechanical when it is small, local, and unambiguous from the diff context (for example: delete unreachable lines; replace `Count() > 0` with `not IsEmpty()`; add a missing `ToolTip`, `OptionCaption`, or `DataClassification`; replace a string-concatenated `Error` with a Label-backed call; change an over-broad permission token; or add an obvious `else`/guard branch). For mechanical findings, emit `findings[].suggested-code` with the literal replacement for the source lines indicated by `location`. The payload must be a verbatim replacement — no diff markers, no fences, no commentary — that the consumer can render as a one-click suggestion. When a `.good.al` companion exists and the diff context matches the `.bad.al` shape, adapt the `.good.al` replacement into `suggested-code`.
 
 Omit `suggested-code` only when the appropriate fix depends on context the skill cannot determine, when multiple defensible replacements exist, or when the fix spans non-contiguous code. If a finding is mechanical-looking but you omit `suggested-code`, set `findings[].suggested-code-omission-reason` to a short explanation. See `skills/do.md` for the full contract.
 

microsoft/skills/review/al-upgrade-review.md

@@ -66,7 +66,7 @@ Set `confidence` to:
 
 After evaluating each worklist entry, also consider whether the diff exhibits a upgrade defect the agent recognises from its general AL knowledge that no knowledge file in the worklist covers. Such candidates are agent findings within this skill's domain — emit them with `references: []`, an `id` slug prefixed with `agent:`, `confidence` capped at `medium`, `severity` capped at `minor` (agent findings are advisory and non-gating), and a `message` that is self-contained (describing both the issue and a concrete recommendation, since there is no knowledge-file footer for the consumer to fall back on). Hold every candidate to the precision bar in `skills/do.md` (*Agent findings*): emit only a concrete, material upgrade or breaking-change defect a knowledgeable BC reviewer would agree is wrong — steelman it first and drop anything stylistic, speculative, dependent on code outside the diff, or merely a valid alternative; when in doubt, omit. The scope is strictly upgrade; defects outside this domain belong to other leaves and MUST NOT be emitted here. Before emitting, check the worklist for a knowledge file that matches the candidate — if one exists, upgrade the candidate to a knowledge-backed finding instead. See `skills/do.md` for the full contract.
 
-For every emitted finding, decide whether the fix is mechanical. A fix is mechanical when it is small, local, and unambiguous from the diff context (for example: delete unreachable lines; replace `Count() > 0` with `not IsEmpty()`; move a local `Label` to object scope; add a missing `ToolTip`, `OptionCaption`, or `DataClassification`; replace a string-concatenated `Error` with a Label-backed call; change an over-broad permission token; or add an obvious `else`/guard branch). For mechanical findings, emit `findings[].suggested-code` with the literal replacement for the source lines indicated by `location`. The payload must be a verbatim replacement — no diff markers, no fences, no commentary — that the consumer can render as a one-click suggestion. When a `.good.al` companion exists and the diff context matches the `.bad.al` shape, adapt the `.good.al` replacement into `suggested-code`.
+For every emitted finding, decide whether the fix is mechanical. A fix is mechanical when it is small, local, and unambiguous from the diff context (for example: delete unreachable lines; replace `Count() > 0` with `not IsEmpty()`; add a missing `ToolTip`, `OptionCaption`, or `DataClassification`; replace a string-concatenated `Error` with a Label-backed call; change an over-broad permission token; or add an obvious `else`/guard branch). For mechanical findings, emit `findings[].suggested-code` with the literal replacement for the source lines indicated by `location`. The payload must be a verbatim replacement — no diff markers, no fences, no commentary — that the consumer can render as a one-click suggestion. When a `.good.al` companion exists and the diff context matches the `.bad.al` shape, adapt the `.good.al` replacement into `suggested-code`.
 
 Omit `suggested-code` only when the appropriate fix depends on context the skill cannot determine, when multiple defensible replacements exist, or when the fix spans non-contiguous code. If a finding is mechanical-looking but you omit `suggested-code`, set `findings[].suggested-code-omission-reason` to a short explanation. See `skills/do.md` for the full contract.