@jasonbelmonti/markdown-engine 3.5.0 → 3.8.0

package/docs/contracts/declarative-validation.md

@@ -1,10 +1,11 @@
 # Declarative Validation Contract
 
 Status: package 3.0.0, v1 profile syntax with v2 Conditional V2, document contract 1.0.0
-Last updated: 2026-08-15
+Last updated: 2026-09-25
 Current v2 surface: flat-rule result/evidence shell, generic selector
 `selectionCount` bounds, document `sourceLength` schema and runtime
-measurement, exact normalized table-header `tableColumnsExact` assertions, ID count-bound schema and
+measurement, opt-in body-row `tableRowsComplete` assertions, exact normalized
+table-header `tableColumnsExact` assertions, ID count-bound schema and
 runtime evaluator contract, plus `tableColumnCoverage` schema, compiled-plan,
 and runtime evaluator contract, `frontmatterShape` schema, compiled-plan, and
 runtime evaluator contract, `textFormat` schema, compiled-plan, and runtime
@@ -237,7 +238,7 @@ emit `profile.config.invalidShape`.
 Rule-level `when` is allowed only on v2 rules. Branch-level `when` remains
 unsupported. V1 profiles preserve the original flat rule authoring contract;
 grouped `anyOf` / `allOf`, ID count bounds, `tableColumnCoverage`,
-`tableColumnsExact`, `frontmatterShape`, `textFormat`, and rule-level `when` are v2 additions.
+`tableColumnsExact`, `tableRowsComplete`, `frontmatterShape`, `textFormat`, and rule-level `when` are v2 additions.
 
 Profile values must be JSON-safe data properties after YAML materialization.
 Functions, accessors, proxies, cyclic structures, sparse arrays, `undefined`
@@ -308,6 +309,7 @@ interface DeclarativeAssertion {
   tableColumnsRequired?: {
     columns: readonly string[];
   };
+  tableRowsComplete?: true; // v2 only
   tableColumnsExact?: {
     columns: readonly string[];
   };
@@ -323,7 +325,9 @@ interface DeclarativeAssertion {
     mustAppearIn: readonly string[];
   };
   tableColumnCoverage?: {
+    allowEmptySource?: boolean;
     source: {
+      rowWhere?: { column: string; equals?: string; includes?: string };
       section: string;
       column: string;
       prefix?: string;
@@ -349,6 +353,7 @@ interface DeclarativeAssertion {
     }[];
   };
   text?: {
+    nonBlank?: true; // v2 only
     contains?: string;
     excludes?: readonly string[];
   };
@@ -389,6 +394,7 @@ Selector/assertion compatibility is part of the public contract:
 | `sectionsRequired` | `document` |
 | `tableColumnsRequired` | `table` |
 | `tableColumnsExact` | `table` |
+| `tableRowsComplete` | `table` |
 | `ids` | all supported selector targets |
 | `references` | `document` |
 | `tableColumnCoverage` | `document` |
@@ -430,6 +436,32 @@ the normal selected-table source evidence is retained. This assertion does not
 change the ordered-subsequence behavior of `tableColumnsRequired` or table
 selector `header` / `tableHeader` matching.
 
+For v2 profiles, `tableRowsComplete: true` requires every body row of each
+selected table to have exactly that table's normalized header column positions.
+Only `true` is admitted; `false`, objects, strings and other payloads emit
+`profile.config.invalidShape`. Only `table` selectors are compatible. V1 rejects
+the assertion through the existing unsupported-assertion/key behavior.
+
+The assertion compares normalized cell coordinates, not text or Markdown pipe
+characters. Missing and excess cells both fail. An explicitly present empty cell
+passes shape; compose `text.nonBlank` on required `tableCell` columns to reject
+blank content. A header-only table passes shape; compose a `tableRow`
+`selectionCount` minimum to require body rows. No matched tables fails with
+`profile.validation.emptySelection`. Existing table section/header selectors
+scope the check, including tables in descendant sections; unselected malformed
+tables do not fail it. Header matching and normalization are unchanged.
+
+Each malformed row emits one `profile.validation.assertionFailed` at the rule's
+severity. Its deterministic message names the table target ID, body row index
+(header is row 0; first body row is 1), and expected/actual zero-based column
+positions. For example: `Selected table "node:0:table" body row 1 must have column
+positions [0,1]; found [0].` Source evidence is the first available body-cell
+range in column order, not a fabricated missing-cell or full-row range. When no
+body-cell range is available, the diagnostic omits `sourceRange`, even if a table
+range is known. Normal diagnostic ordering applies; without ranges, selected-table
+order and numeric row order are retained. This check needs no original source
+text and does not alter GFM parsing or strengthen existing profiles.
+
 `sectionsRequired.order` defaults to `none`. `strict` checks that configured
 headings appear as an ordered subsequence in the normalized section tree
 flattened in source order.
@@ -461,6 +493,22 @@ target section do not satisfy coverage. Missing target sections, missing target
 columns, and missing target-column IDs emit deterministic validation diagnostics
 source-grounded to the source ID when source evidence is available.
 
+The opt-in `source.rowWhere` filter uses the same closed sibling-cell predicate
+as the `tableCell` selector: a non-empty `column` plus `equals`, `includes`, or
+both. Both predicates must match when both are present. Matching uses normalized,
+case-sensitive cell text before ID extraction; `source.caseSensitive` affects ID
+comparison only. Rows marked optional can therefore be excluded without encoding
+consumer vocabulary in the engine. A missing predicate column is unresolved,
+not a resolved empty source.
+
+`allowEmptySource` is an optional boolean. Absent or `false` preserves the existing
+`profile.validation.emptySelection` failure when no source IDs are extracted.
+`true` permits zero extracted IDs only when the source section and column resolve
+(including the predicate column when configured). With no source IDs there are no
+target obligations; target resolution is not evaluated. A missing source structure
+still fails. Pair this policy with consumer shape rules when tables or rows are
+mandatory. This option also permits prose sentinels in reverse declared-ID checks.
+
 For v2 profiles, `frontmatterShape` is admitted as a flat-rule schema and
 internal compiled-plan assertion. It is compatible only with a `document`
 selector because frontmatter is document metadata. `presence` is optional and
@@ -505,7 +553,18 @@ is combined with `equals` or `nonBlank`, a non-string value emits only
 string-predicate diagnostic. Diagnostic messages do not include frontmatter
 values.
 
-`text` must include `contains` or a non-empty `excludes` array.
+`text` must include `contains`, a non-empty `excludes` array, or (v2 only)
+`nonBlank: true`. `nonBlank` cannot be `false`. It requires at least one character
+after JavaScript `trim()` on normalized node text excluding raw `html`, `definition`,
+and `yaml` nodes. Comments, empty HTML tags, and decoded whitespace entities cannot
+satisfy it. Markdown emphasis, link labels, literal code, and escaped angle brackets
+retain their text. Each selected target is checked independently; an empty selector
+still fails. Section selection includes its heading and direct body, as with ordinary
+section text. Original source ranges are optional; the normalized nodes provide the
+content. This is a structural textual-content check, not CSS visibility evaluation
+or browser rendering. It does not evaluate external evidence, freshness, truth,
+or implementation acceptance. Existing `contains`, `excludes`, and `textLength`
+semantics are unchanged, even when combined with `nonBlank`.
 `textOccurrenceCount.count` is a finite number and counts non-overlapping
 literal occurrences per selected target.
 `textLength` must include `min`, `max`, or both. Bounds are non-negative
@@ -538,7 +597,7 @@ parsing, or implement date ordering.
 
 Empty selector results produce `profile.validation.emptySelection` for exists,
 table, ID, reference, text, occurrence, text-length, text-format, and
-tableColumnsExact assertions.
+tableColumnsExact and tableRowsComplete assertions.
 `selectionCount` instead evaluates the empty selection as zero. Document-scoped
 required-section, required-frontmatter, and frontmatter-shape assertions
 evaluate against the document.
@@ -592,6 +651,7 @@ an unproven pass.
 | `profile.validation.tableColumnCoverageIdMissing` | Rule severity | A source ID is absent from the configured target table column. |
 | `profile.validation.tableColumnCoverageTargetColumnMissing` | Rule severity | The configured target table column cannot be resolved. |
 | `profile.validation.tableColumnCoverageTargetSectionMissing` | Rule severity | The configured target section cannot be resolved. |
+| `profile.validation.textBlank` | Rule severity | A target configured with `text.nonBlank` lacks non-whitespace text outside raw HTML. |
 | `profile.validation.textExcluded` | Rule severity | A selected target contains forbidden literal text. |
 | `profile.validation.textMissing` | Rule severity | A selected target lacks required literal text from a `text.contains` assertion. |
 | `profile.validation.assertionUnsupported` | `error` | A compiled assertion or compiled assertion feature has no evaluator implementation; this is an internal safety diagnostic. |
@@ -738,7 +798,7 @@ that use `sourceLength`.
 The declarative validation CLI command is:
 
 ```sh
-markdown-engine validate --file <markdown-file> --profile <profile-file> [--format json]
+markdown-engine validate --file <markdown-file> --profile <profile-file> [--format json] [--output full|summary] [--report-file <new-file>]
 ```
 
 `--format json` is the default and only supported validation output format.
@@ -757,7 +817,7 @@ union; there is no extra CLI discriminator beyond
 
 ## CLI JSON Union
 
-The CLI JSON output is:
+The default (`--output full`) CLI JSON output is:
 
 ```ts
 type DeclarativeValidationCliJsonResult =
@@ -785,13 +845,70 @@ profiles, that same validation-result JSON can include `evaluatedRuleCount`,
 `skippedRuleCount`, `status: "skipped"`, nested `when` diagnostics, and
 `evaluation.kind: "skipped"`.
 
+## Compact Validation Output
+
+`--output full` (the default) preserves the existing full JSON contract and
+serialization. `--output summary --report-file <new-file>` selects a separate
+CLI-only `schemaVersion: "markdown-engine.validation-summary.v1"` representation.
+Both modes still use `--format json`. Each new selector accepts spaced or
+assignment syntax and may occur only once. Missing/blank paths, unsupported
+output modes and summary mode without a report path are usage errors (exit 2).
+These options do not apply to the normalization command.
+
+Summary fields:
+
+| Field | Meaning |
+| --- | --- |
+| `schemaVersion` | Exact `markdown-engine.validation-summary.v1` discriminator. |
+| `valid`, `exitCode` | Original validation verdict and its 0/1 CLI status. |
+| `stage` | `profile` for rejected profiles; otherwise `validation`, including normalization failures. |
+| `engineVersion`, `runtimeVersion` | Producing package and Node versions, including profile-stage failures. |
+| `profile` | Existing admitted-profile metadata, when available, including V2 evaluated/skipped counts. |
+| `evidence` | Existing input/profile hashes, engine/runtime versions and optional sourceLength; no duplicated rule results or diagnostics. Omitted when full-result evidence is absent. |
+| `diagnosticCounts` | Total and error/warning/info counts over all top-level diagnostics. |
+| `diagnostics` | Up to ten top-level diagnostics, ordered error, warning, info; existing order is preserved within a severity. |
+| `diagnosticsOmitted` | Number of top-level diagnostics not shown. |
+| `diagnosticsTruncated` | Number of shown diagnostics with at least one shortened field. |
+| `report` | Absolute `path`, UTF-8 `bytes`, and raw-file `sha256` of the complete retained result, including its final newline. |
+
+Shown diagnostics preserve severity and available sourceRange. Messages are
+limited to 512 UTF-16 code units; code and ruleId to 128 each, including a final
+ellipsis when shortened. Each diagnostic has `truncatedFields`, naming exactly
+which fields were shortened (empty when none). Full unmodified fields remain
+in the report. Counts exclude nested skipped-applicability and branch details,
+which remain in the complete result. A passing verdict can contain warnings;
+summary presentation neither discards those warnings nor changes the exit code.
+
+The full report is byte-identical to default stdout for the same invocation's
+validation result: stable pretty JSON followed by one newline. Its raw-file
+SHA-256 is distinct from normalized `evidence.inputHash` and `profileHash`.
+Summary stdout is stable compact JSON followed by one newline, with no rule
+result arrays. These presentation fields are not additions to the engine's
+public API result types or evidence hashes.
+
+`--report-file` is also permitted with full output. Relative destinations resolve
+against the invocation's current working directory. The parent directory must
+exist; the destination must not exist. Report publication stages a complete file
+in that directory, then publishes it with an exclusive hard link and removes
+the temporary file. Existing files, directories, symlinks and hard links are
+never replaced. Filesystems that do not support this operation report an I/O
+error. Report bytes are written before any validation JSON is emitted to stdout.
+
+A report-publication failure overrides a 0/1 validation status with exit 2,
+emits a stderr error, and emits no validation JSON to stdout. Usage or input-read
+failures continue to use the existing stderr contract and do not create a report.
+Profile-stage validation failures do create a full report when requested, retain
+exit 1, and omit unavailable evidence identities from the summary. The report
+writer is a narrow CLI filesystem-output boundary; the validator/API do not gain
+persistence or filesystem-write behavior.
+
 ## Exit Codes
 
 | Exit code | Meaning |
 | --- | --- |
 | `0` | Validation completed with no top-level error-severity diagnostics. |
 | `1` | Profile/config/compile, Markdown normalization, document-version mismatch, or top-level validation diagnostics include at least one error. |
-| `2` | CLI usage, unsupported format, unknown argument, missing argument value, repeated singleton flag, unsupported `--document-version`, or local file read error. |
+| `2` | CLI usage, unsupported format, unknown argument, missing argument value, repeated singleton flag, unsupported `--document-version`, or local file read/report-publication error. |
 
 ## Compatibility And Migration
 
@@ -896,6 +1013,36 @@ Migration notes:
 
 ## Examples
 
+### Opt-in table body completeness
+
+```yaml
+syntaxVersion: markdown-engine.validation@v2
+documentVersion: 1.0.0
+rules:
+  - id: sources.rows.complete
+    select:
+      target: table
+      section: Sources and baseline
+    assert:
+      tableRowsComplete: true
+  - id: sources.identity.nonblank
+    select:
+      target: tableCell
+      section: Sources and baseline
+      column: SHA-256 / immutable identity
+    assert:
+      text:
+        nonBlank: true
+```
+
+The [runnable table-row example](../../fixtures/declarative-validation/examples/table-rows-complete/README.md)
+adapts delegation-planner's missing-dynamic-row-cell reproducer. It demonstrates
+legacy header-only validation passing the truncated row, opt-in shape validation
+failing at that row, and a repaired document passing directly through Engine.
+Consumer-owned section selection remains profile policy; no supplemental shape
+parser is needed. Older installed runtimes do not admit this new assertion;
+adoption requires a separately released compatible runtime.
+
 Minimal profile:
 
 ```yaml

package/fixtures/declarative-validation/examples/table-rows-complete/README.md

@@ -0,0 +1,65 @@
+# Table body completeness
+
+`tableRowsComplete: true` is an opt-in v2 assertion for `table` selectors. It
+compares each body row's normalized column positions with its table's header.
+Missing and excess cells fail. Header validation remains independent.
+
+From the repository root, build and run the example:
+
+```sh
+npm run build
+node dist/cli/index.js validate --file fixtures/declarative-validation/examples/table-rows-complete/consumer-missing.md --profile fixtures/declarative-validation/examples/table-rows-complete/legacy.yaml --format json
+node dist/cli/index.js validate --file fixtures/declarative-validation/examples/table-rows-complete/consumer-missing.md --profile fixtures/declarative-validation/examples/table-rows-complete/profile.yaml --format json
+node dist/cli/index.js validate --file fixtures/declarative-validation/examples/table-rows-complete/consumer-repaired.md --profile fixtures/declarative-validation/examples/table-rows-complete/profile.yaml --format json
+```
+
+Expected exits are **0, 1, 0**. The second command reports rule `sources.columns`,
+body row 2 at source line 20, expected `[0,1,2,3]`, actual `[0,1,2]`. Its range
+locates the first available cell of the failing row. Table and row identities
+remain in the message when cell ranges are unavailable; no location is invented.
+
+`consumer-missing.md` is adapted from delegation-planner commit
+`bc11ace3e50315313d428f3b681424fee910aef3`,
+`skills/delegation-planner/validation/fixtures/missing-dynamic-row-cell.md`,
+with an unrelated metadata row omitted.
+`consumer-repaired.md` adds the missing execution-plan identity cell.
+`legacy.yaml` checks the exact source-table header; `profile.yaml` adds only the
+row-completeness assertion. This focused adaptation proves Engine's shape decision
+without invoking or migrating the consumer helper. It does not claim to validate
+all delegation policy, source integrity, reserved markers or semantic readiness.
+
+`oracle.json` records independently authored valid/invalid decisions, row indexes,
+source lines and actual column positions for the small Markdown fixtures. API and
+real CLI tests use these expectations. Fixtures include escaped pipes, inline code,
+Unicode, a blockquote table, excess and truncated rows, and explicit empty cells.
+
+An explicit empty cell (`| x | |`) passes shape. Require nonblank values separately:
+
+```yaml
+- id: values.required
+  select:
+    target: tableCell
+    column: B
+  assert:
+    text:
+      nonBlank: true
+```
+
+A header-only table also passes shape. To require body rows, compose:
+
+```yaml
+- id: rows.required
+  select:
+    target: tableRow
+  assert:
+    selectionCount:
+      min: 1
+```
+
+Apply the same section/header scope to composed rules when appropriate. Unmatched
+table selection fails with `profile.validation.emptySelection`; unselected
+malformed tables do not affect the rule. Section selection includes descendant
+sections. The Engine parser and normalized document remain unchanged.
+
+Older runtimes reject this assertion. These commands use the local build;
+release, installed-runtime adoption and consumer-helper removal are separate work.

package/fixtures/declarative-validation/examples/table-rows-complete/blockquote.md

@@ -0,0 +1,3 @@
+> | A | B |
+> | --- | --- |
+> | x |

package/fixtures/declarative-validation/examples/table-rows-complete/complete.md

@@ -0,0 +1,3 @@
+| A | B |
+| --- | --- |
+| x | y |

package/fixtures/declarative-validation/examples/table-rows-complete/consumer-missing.md

@@ -0,0 +1,50 @@
+---
+type: DelegationPlan
+artifact_version: "2.0"
+revision: "1"
+delegation_id: fixture
+---
+# Structural fixture only
+
+## Control
+
+| State | Granularity | Selected pilot extent | Max attempts per packet | Scheduling |
+| --- | --- | --- | --- | --- |
+| READY | action | DP-WP-1 | 2 | Serial |
+
+## Sources and baseline
+
+| Role | Full read path / reference | Revision | SHA-256 / immutable identity |
+| --- | --- | --- | --- |
+| Source contract | /fixture/task.md | 1 | synthetic-source-identity |
+| Execution plan | /fixture/plan.md | 1 |
+| Repository baseline and instructions | /fixture/repository | initial | synthetic-code-identity |
+| Upstream validation | /fixture/validation.json | 1 | synthetic-validation-identity |
+
+## Model roster
+
+| Order | Model | Reasoning effort | Harness | Availability evidence |
+| --- | --- | --- | --- | --- |
+| 1 | gpt-6-luna | medium | manual | Unverified; check before dispatch |
+
+## Ordered assignments
+
+| Packet / coordinator | Ordered source steps | Source outcomes | Worker packet | Model / effort | Selection basis | Escalation / stop |
+| --- | --- | --- | --- | --- | --- | --- |
+| DP-WP-1 | EP-ACT-1, EP-GATE-1 | TD-SC-1 | /fixture/DP-WP-1.md | gpt-6-luna / medium | experimental; bounded fixture | Stop after two total attempts |
+
+## Readiness
+
+| Field | Value |
+| --- | --- |
+| Audit decision | PASS |
+| Audit evidence | Synthetic mapping only; not dispatchable real work |
+| Dispatch prerequisites | Obtain real sources, authorization and model availability |
+| Blockers | None |
+| Resume condition | Not applicable |
+
+## Revision note
+
+| Field | Value |
+| --- | --- |
+| Change and authority | Initial independent structural fixture; no implementation authority |

package/fixtures/declarative-validation/examples/table-rows-complete/consumer-repaired.md

@@ -0,0 +1,50 @@
+---
+type: DelegationPlan
+artifact_version: "2.0"
+revision: "1"
+delegation_id: fixture
+---
+# Structural fixture only
+
+## Control
+
+| State | Granularity | Selected pilot extent | Max attempts per packet | Scheduling |
+| --- | --- | --- | --- | --- |
+| READY | action | DP-WP-1 | 2 | Serial |
+
+## Sources and baseline
+
+| Role | Full read path / reference | Revision | SHA-256 / immutable identity |
+| --- | --- | --- | --- |
+| Source contract | /fixture/task.md | 1 | synthetic-source-identity |
+| Execution plan | /fixture/plan.md | 1 | synthetic-plan-identity |
+| Repository baseline and instructions | /fixture/repository | initial | synthetic-code-identity |
+| Upstream validation | /fixture/validation.json | 1 | synthetic-validation-identity |
+
+## Model roster
+
+| Order | Model | Reasoning effort | Harness | Availability evidence |
+| --- | --- | --- | --- | --- |
+| 1 | gpt-6-luna | medium | manual | Unverified; check before dispatch |
+
+## Ordered assignments
+
+| Packet / coordinator | Ordered source steps | Source outcomes | Worker packet | Model / effort | Selection basis | Escalation / stop |
+| --- | --- | --- | --- | --- | --- | --- |
+| DP-WP-1 | EP-ACT-1, EP-GATE-1 | TD-SC-1 | /fixture/DP-WP-1.md | gpt-6-luna / medium | experimental; bounded fixture | Stop after two total attempts |
+
+## Readiness
+
+| Field | Value |
+| --- | --- |
+| Audit decision | PASS |
+| Audit evidence | Synthetic mapping only; not dispatchable real work |
+| Dispatch prerequisites | Obtain real sources, authorization and model availability |
+| Blockers | None |
+| Resume condition | Not applicable |
+
+## Revision note
+
+| Field | Value |
+| --- | --- |
+| Change and authority | Initial independent structural fixture; no implementation authority |

package/fixtures/declarative-validation/examples/table-rows-complete/excess.md

@@ -0,0 +1,3 @@
+| A | B |
+| --- | --- |
+| x | y | z |

package/fixtures/declarative-validation/examples/table-rows-complete/explicit-empty.md

@@ -0,0 +1,3 @@
+| A | B |
+| --- | --- |
+| x | |

package/fixtures/declarative-validation/examples/table-rows-complete/header-only.md

@@ -0,0 +1,2 @@
+| A | B |
+| --- | --- |

package/fixtures/declarative-validation/examples/table-rows-complete/legacy.yaml

@@ -0,0 +1,10 @@
+syntaxVersion: markdown-engine.validation@v2
+documentVersion: 1.0.0
+rules:
+  - id: sources.columns
+    select:
+      target: table
+      section: Sources and baseline
+    assert:
+      tableColumnsExact:
+        columns: [Role, Full read path / reference, Revision, SHA-256 / immutable identity]

package/fixtures/declarative-validation/examples/table-rows-complete/missing.md

@@ -0,0 +1,3 @@
+| A | B |
+| --- | --- |
+| x |

package/fixtures/declarative-validation/examples/table-rows-complete/multiple.md

@@ -0,0 +1,5 @@
+| A | B |
+| --- | --- |
+| x |
+| x | y | z |
+| z | q |

package/fixtures/declarative-validation/examples/table-rows-complete/oracle.json

@@ -0,0 +1,85 @@
+[
+  {
+    "name": "complete",
+    "valid": true,
+    "failures": []
+  },
+  {
+    "name": "missing",
+    "valid": false,
+    "failures": [
+      {
+        "row": 1,
+        "line": 3,
+        "actual": [
+          0
+        ]
+      }
+    ]
+  },
+  {
+    "name": "excess",
+    "valid": false,
+    "failures": [
+      {
+        "row": 1,
+        "line": 3,
+        "actual": [
+          0,
+          1,
+          2
+        ]
+      }
+    ]
+  },
+  {
+    "name": "explicit-empty",
+    "valid": true,
+    "failures": []
+  },
+  {
+    "name": "header-only",
+    "valid": true,
+    "failures": []
+  },
+  {
+    "name": "syntax",
+    "valid": true,
+    "failures": []
+  },
+  {
+    "name": "multiple",
+    "valid": false,
+    "failures": [
+      {
+        "row": 1,
+        "line": 3,
+        "actual": [
+          0
+        ]
+      },
+      {
+        "row": 2,
+        "line": 4,
+        "actual": [
+          0,
+          1,
+          2
+        ]
+      }
+    ]
+  },
+  {
+    "name": "blockquote",
+    "valid": false,
+    "failures": [
+      {
+        "row": 1,
+        "line": 3,
+        "actual": [
+          0
+        ]
+      }
+    ]
+  }
+]

package/fixtures/declarative-validation/examples/table-rows-complete/profile.yaml

@@ -0,0 +1,11 @@
+syntaxVersion: markdown-engine.validation@v2
+documentVersion: 1.0.0
+rules:
+  - id: sources.columns
+    select:
+      target: table
+      section: Sources and baseline
+    assert:
+      tableColumnsExact:
+        columns: [Role, Full read path / reference, Revision, SHA-256 / immutable identity]
+      tableRowsComplete: true

package/fixtures/declarative-validation/examples/table-rows-complete/shape.yaml

@@ -0,0 +1,8 @@
+syntaxVersion: markdown-engine.validation@v2
+documentVersion: 1.0.0
+rules:
+  - id: rows.complete
+    select:
+      target: table
+    assert:
+      tableRowsComplete: true

package/fixtures/declarative-validation/examples/table-rows-complete/syntax.md

@@ -0,0 +1,3 @@
+| A | B |
+| --- | --- |
+| 雪 \| 🌲 | `a \| b` |

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@jasonbelmonti/markdown-engine",
-  "version": "3.5.0",
+  "version": "3.8.0",
   "description": "Deterministic Markdown parsing and validation engine package.",
   "license": "MIT",
   "type": "module",
@@ -71,7 +71,7 @@
     "test:validation:profile": "npm run build && vitest run tests/declarative-validation-profile.test.ts \"--exclude=.worktrees/**\"",
     "test:validation:compiler": "npm run build && vitest run tests/declarative-validation-compiler.test.ts \"--exclude=.worktrees/**\"",
     "test:validation:selectors": "npm run build && vitest run tests/declarative-validation-selectors.test.ts \"--exclude=.worktrees/**\"",
-    "test:validation:assertions": "npm run build && vitest run tests/declarative-validation-assertions.test.ts tests/declarative-validation-frontmatter-shape-assertions.test.ts tests/declarative-validation-table-column-coverage-fixtures.test.ts tests/declarative-validation-table-columns-exact.test.ts tests/declarative-validation-source-assertion-coverage.test.ts tests/declarative-validation-source-length.test.ts tests/declarative-validation-selection-count.test.ts tests/declarative-validation-normalized-source-ranges.test.ts tests/declarative-validation-grouped-rules-fixtures.test.ts tests/declarative-validation-when-skipped-rules-fixtures.test.ts \"--exclude=.worktrees/**\"",
+    "test:validation:assertions": "npm run build && vitest run tests/declarative-validation-assertions.test.ts tests/declarative-validation-filtered-coverage.test.ts tests/declarative-validation-visible-text.test.ts tests/declarative-validation-frontmatter-shape-assertions.test.ts tests/declarative-validation-table-column-coverage-fixtures.test.ts tests/declarative-validation-table-columns-exact.test.ts tests/declarative-validation-table-rows-complete.test.ts tests/declarative-validation-table-rows-complete-config.test.ts tests/declarative-validation-table-rows-complete-cli.test.ts tests/declarative-validation-source-assertion-coverage.test.ts tests/declarative-validation-source-length.test.ts tests/declarative-validation-selection-count.test.ts tests/declarative-validation-normalized-source-ranges.test.ts tests/declarative-validation-grouped-rules-fixtures.test.ts tests/declarative-validation-when-skipped-rules-fixtures.test.ts \"--exclude=.worktrees/**\"",
     "test:validation:diagnostics": "npm run build && vitest run tests/declarative-validation-diagnostics.test.ts \"--exclude=.worktrees/**\"",
     "test:validation:cli": "npm run build && vitest run tests/declarative-validation-cli.test.ts \"--exclude=.worktrees/**\"",
     "test:validation:examples": "npm run build && vitest run tests/declarative-validation-examples.test.ts \"--exclude=.worktrees/**\"",

package/scripts/install-markdown-engine-cli.sh

@@ -1,9 +1,9 @@
 #!/usr/bin/env sh
 set -eu
 
-VERSION="3.5.0"
+VERSION="3.8.0"
 PACKAGE="@jasonbelmonti/markdown-engine"
-EXPECTED_SHA256="4a4e9ae7d9ff4797c96a7483555bcba7cbec5852d330c6c50dc0ef8fb2e5d933"
+EXPECTED_SHA256="cecfb88ab9cb9a3030c13e41b5e56ac30b44436fb1da7bb890bea5f9bd215fce"
 DEFAULT_DATA_HOME="${XDG_DATA_HOME:-$HOME/.local/share}"
 MARKDOWN_ENGINE_HOME="${MARKDOWN_ENGINE_HOME:-$DEFAULT_DATA_HOME/markdown-engine}"
 MARKDOWN_ENGINE_BIN_DIR="${MARKDOWN_ENGINE_BIN_DIR:-$HOME/.local/bin}"