chore: prepare Methodology 6.0.0 release

CHANGELOG.md

@@ -8,6 +8,71 @@ Format follows [Keep a Changelog](https://keepachangelog.com/en/1.1.0/). Version
 
 ## [Unreleased]
 
+## [6.0.0] — Unreleased
+
+Bump category: **MAJOR**. Knowledge supersession now enforces errors on admitted
+objects that the 5.1.0 linter accepted, including the published `KS-019`
+warning-to-error change. Follow the [5.1 → 6.0 migration recipe](migrations/5.1-to-6.0/)
+before upgrading. The manifest and ingest coverage presets both target 6.0.0;
+package versions retain their independent versioning.
+
+### Added
+
+- Repository MCP pattern and enterprise setup guidance, including responsibilities
+  for catalogue hosts and consumers. (#587, #603)
+- Catalogue-release and vocabulary CI examples, plus ingest guidance for
+  recognition, promotion, and review of returned bindings. (#592, #594)
+- HTML glossary reports derived from admitted canon through the document-view
+  engine's `view` directive. (#593)
+- Cold-read adoption review procedure and expanded knowledge freshness tests. (#588, #595)
+
+### Changed
+
+- **Breaking: knowledge supersession validation.** `KS-018` checks both pointers
+  for a unique admitted target; `KS-019` requires reciprocal admitted pointers;
+  `KS-020` rejects mismatches, self-links and cycles. A draft proposing
+  `supersedes` still receives only a missing-backlink warning. Linear chains
+  may grow beyond two objects. Re-curation preserves original assertions and
+  sources, admits a successor, and updates only predecessor lifecycle metadata.
+  The knowledge-store skill minimum version is 6.0.0. (#589, #597)
+- **Extraction compatibility:** unsupported top-level extraction fields now fail
+  explicitly instead of disappearing; schema-undefined data belongs in
+  `extensions`. Corroboration preserves existing values while filling missing
+  fields and adding provenance. Review custom extraction producers before
+  upgrading. (#602)
+- Validator coverage contract requires exhaustive file accounting, visible
+  unvalidated paths and warnings, and `NOTATION-SKIP-001` promotion under
+  `--strict`. Action scheduling advisories are documented at warning severity.
+  Downstream tool conformance must be checked separately. (#599)
+- Compaction guidance requires demonstrated reachability and preservation;
+  base/open-tier stores defer removal. No compaction algorithm ships. (#598)
+- Issued-document preservation and assembly-history guidance: each release records
+  its full environment, earlier assembly facts remain available, and RELEASE
+  subjects remain PRODUCT or APPLICATION. No new release-state schema is added. (#600)
+
+### Fixed
+
+- Candidate TYPE fields survive extraction and review without importing source
+  claims of admission authority. (#602)
+- The first Goals-tree session and one-command L0 adoption instructions are
+  shorter and directly executable. (#590, #591)
+- Plugin example dependency `js-yaml` updates from 4.3.1 to 4.3.2. (#596)
+- Knowledge-store rule tables now match enforced supersession severity and scope.
+
+### Removed
+
+- Obsolete documentation audit report. (#586)
+
+### Upgrade compatibility
+
+Stores without supersession pointers need no knowledge-object rewrite. Repair
+existing links through review; never auto-create a relationship or mutate an
+admitted assertion to satisfy validation. Refresh the knowledge-store linter
+and skill alongside the manifest, specs, and ingest CLI. `tools/lint.py` is
+unchanged. Coverage preset element and relation lists are unchanged. Independent
+notation revisions updated for this span are Action 2.3, Relations 0.4 and
+Glossary 0.2; other notation edits only refresh example methodology pins.
+
 ## [5.1.0] — 2026-09-04
 
 Bump category: **MINOR** — highest change since the 5.0.0 pin is additive: the `documents` package, knowledge-object supersession fields and freshness warnings, the issue-tracker binding guide, and document-renderer paged-media / landscape / caption work. No migration recipe: a repository that never declares `packages: [documents]` and never uses the new knowledge-store fields validates as it did on 5.0.0. New `error` codes (`DOCS-001..007`, `KS-018`, `KS-020`) fire only when those optional surfaces are in use.

integration/test_catalogue_workflows.py

@@ -69,7 +69,7 @@ def test_publication_rejects_mismatch_extra_fields_duplicates_and_lossy_yaml(sel
 
     def test_ci_reports_both_findings_without_editing_canon_and_fails_on_broken_pin(self):
         manifest = self.root / 'transitrix.yaml'
-        manifest.write_text('methodology_version: "5.1.0"\n')
+        manifest.write_text('methodology_version: "6.0.0"\n')
         result = self.run_step('ci-example.yaml', 'vocabulary', 'Report collisions and unbound matches')
         self.assertEqual((result.returncode, result.stdout), (0, ''))
         (self.root / 'vendor').mkdir()

method/06-team-operations.md

@@ -110,7 +110,7 @@ Deliberately **not** one-file-per-record like ADR/WI: the journal is a single fi
 ---
 ### FB-0002
 type: notation-gap
-methodology_version: "5.1.0"
+methodology_version: "6.0.0"
 raised_by: modeler
 date: "2026-07-27"
 status: sent-upstream

migrations/5.1-to-6.0/README.md

@@ -0,0 +1,101 @@
+# Migration recipe — methodology 5.1 → 6.0
+
+Version 6.0.0 preserves knowledge supersession history and enforces reciprocal
+links. The v5.1.0 rule table described `KS-019` as a warning; it is now an error
+for admitted objects. `KS-018` and `KS-020` also enforce target resolution and
+graph integrity. A store that passed the old linter can fail the new one.
+Stores without supersession fields need no knowledge-object edits.
+
+| Old form or behaviour | 6.0.0 requirement |
+|---|---|
+| Admitted successor names a predecessor without a backlink | Both admitted objects name each other; `KS-019` is an error |
+| Malformed, missing, ambiguous, or draft target | Each pointer is one non-empty string resolving to one admitted knowledge object (`KS-018`) |
+| Mismatched reverse pointer, self-link, or cycle | Repair the reviewed linear history (`KS-020`); chains of any length are allowed |
+| Draft proposes a successor | Missing predecessor backlink remains a `KS-019` warning; do not mutate the predecessor before approval |
+| Extraction carries unsupported top-level fields | Put schema-undefined data under `extensions`; supported TYPE fields stay at their defined locations |
+| A later extraction overwrites an existing candidate value | Corroboration preserves existing values, fills absent fields, and adds provenance |
+
+## Manual steps
+
+1. Preserve the current revision and review the upgrade in a separate change.
+   Run the **6.0.0** `tools/knowledge_store_lint.py` against the adopter root.
+   Resolve `KS-018`/`KS-020` findings from evidence; do not infer a relationship
+   from filenames or rewrite an assertion to make the linter pass.
+2. For an already approved successor, add the missing reciprocal lifecycle
+   pointer and advance the predecessor's `timestamp` to the repair time. For example:
+
+   ```diff
+   # knowledge/old.md (predecessor)
+   -timestamp: 2026-09-01T10:00:00Z
+   +timestamp: 2026-09-18T10:00:00Z
+   +superseded_by: /knowledge/new.md
+   # knowledge/new.md already contains:
+    supersedes: /knowledge/old.md
+   ```
+
+   Preserve the predecessor's body, `source`, `description`, `confidence`,
+   `created_at`, and any existing `supersedes`. Retain both objects, original
+   sources and hashes, index rows, and a dated admission/repair log. A draft
+   stays under review until the successor and backlink can be admitted together.
+   This requires curator review; the codemod deliberately cannot do it.
+3. Review extraction producers against the supported fields in
+   [`emit-candidates.mjs`](../../packages/ingest-cli/src/emit-candidates.mjs).
+   For example, replace top-level `custom_score: 7` with
+   `extensions: {custom_score: 7}` when that field is not in the schema.
+   Re-extract previously dropped TYPE fields from preserved sources, then review
+   them; do not overwrite reviewed candidates merely to refresh confidence.
+4. Review assembly history against
+   [`17-relations.md`](../../notations/elements/17-relations.md).
+   Each release records its complete `assembled_on` environment, including
+   unchanged systems. A successor does not end its predecessor's historical
+   links. Correct a `valid_to` set solely for succession only after reviewing
+   the original evidence. `RELEASE` subjects are `PRODUCT` or `APPLICATION`;
+   the former `TECHNOLOGY_SERVICE` wording did not override that restriction.
+5. Set `methodology_version` in `transitrix.yaml` from `5.1.0` to `6.0.0`,
+   manually or with the codemod below. Reinstall the ingest CLI from the new
+   checkout, refresh adopted specs and skills, and refresh the knowledge-store
+   linter if vendored. The knowledge-store skill now has `min_version: "6.0.0"`;
+   update the corresponding skill-table entry. `tools/lint.py` itself did not
+   change in this release. Follow the four-artefact procedure in
+   [`RELEASING.md`](../../RELEASING.md#adopter-upgrade-procedure).
+6. Run the post-check below, normal canon validation and `repo-check`. Review
+   warnings explicitly. Validator implementers must also check CONTRACT §18's
+   exhaustive coverage accounting and `NOTATION-SKIP-001` strict promotion;
+   the specification does not certify every downstream validator's support.
+
+Compaction remains a decision frame, with no deletion tool. Retain knowledge
+objects until mechanical reachability and preservation can be demonstrated;
+base/open-tier stores cannot establish the required canon binding and must
+defer removal. See [the pattern](../../patterns/knowledge-store.md).
+
+## Commands and boundaries
+
+Use Node.js ≥20 for the pin-only, dependency-free codemod. The post-check also
+requires `python3` and PyYAML, as does the reference knowledge-store linter.
+Run these commands from the 6.0.0 methodology checkout:
+
+```bash
+node migrations/5.1-to-6.0/codemod.mjs --dry-run <adopter-root>
+node migrations/5.1-to-6.0/codemod.mjs <adopter-root>
+node migrations/5.1-to-6.0/validate.mjs <adopter-root>
+```
+
+The codemod changes only one unambiguous top-level manifest pin and preserves
+other text, comments, quoting and line endings. Rerunning is a no-op. It exits
+`1` for an unsupported/ambiguous pin, `2` for an I/O or invocation error, and
+`0` on success; `--dry-run` writes nothing. Its success is not evidence that
+manual compatibility work is complete. Upgrade older versions through their
+intervening recipes first. For a knowledge-only store without a manifest, use
+the manual steps and run `python3 tools/knowledge_store_lint.py <store-root>`
+directly rather than adding a manifest solely for this wrapper.
+
+`validate.mjs` checks the target pin and invokes the same reference linter used
+by the skill. It exits `0` with no errors, `1` for validation findings, or `2`
+for invocation/I/O errors. It never applies the codemod. It cannot verify
+historical preservation, human approval, extraction producers, or downstream
+validator implementations; those remain the manual review above.
+
+`fixtures/before/` and `fixtures/after/` demonstrate the automated pin change
+on an already valid reciprocal history. The test also removes a backlink to
+prove the post-check rejects incomplete admission; it verifies dry-run,
+idempotence, and byte preservation of the knowledge objects.

migrations/5.1-to-6.0/codemod.mjs

@@ -0,0 +1,29 @@
+#!/usr/bin/env node
+// Only the manifest pin is mechanical. Knowledge relationships require review.
+import { readFileSync, writeFileSync } from 'node:fs';
+import { join, resolve } from 'node:path';
+
+const args = process.argv.slice(2);
+if (args.some(a => a.startsWith('--') && a !== '--dry-run') || args.filter(a => !a.startsWith('--')).length > 1) {
+  console.error('Usage: node codemod.mjs [--dry-run] [target-dir]');
+  process.exit(2);
+}
+try {
+  const file = join(resolve(args.find(a => !a.startsWith('--')) ?? '.'), 'transitrix.yaml');
+  const text = readFileSync(file, 'utf8');
+  const lines = text.match(/^methodology_version:.*$/gm) ?? [];
+  const pin = /^methodology_version: (["']?)(5\.1\.0|6\.0\.0)\1([ \t]*(?:#.*)?)(\r?)$/;
+  if (lines.length !== 1 || !pin.test(lines[0])) {
+    console.error('Manual intervention: expected one top-level 5.1.0 or 6.0.0 methodology_version pin.');
+    process.exit(1);
+  }
+  const next = text.replace(/^methodology_version:.*$/m, line => line.replace(pin, (_, quote, version, comment, cr) => `methodology_version: ${quote}6.0.0${quote}${comment}${cr}`));
+  const changed = next !== text;
+  if (changed && !args.includes('--dry-run')) writeFileSync(file, next);
+  if (changed) console.log('transitrix.yaml: methodology_version 5.1.0 -> 6.0.0 (1 field)');
+  console.log(`1 file scanned; ${changed ? 1 : 0} file${changed ? '' : 's'} ${args.includes('--dry-run') ? 'would change' : 'changed'}.`);
+  console.log('Pin update only: complete the manual compatibility review and run validate.mjs.');
+} catch (error) {
+  console.error(error.message);
+  process.exit(2);
+}

migrations/5.1-to-6.0/fixtures/after/_intake/processed/2026-07-01-cfo-interview.md

@@ -0,0 +1,16 @@
+---
+type: source-document
+title: "CFO onboarding interview"
+description: "First-hand account of the order-fulfilment pain points from the CFO."
+source: "_intake/originals/2026-07-01-cfo-interview.md"
+source_hash: "sha256:0000000000000000000000000000000000000000000000000000000000000001"
+created_at: 2026-07-01
+confidence: observed
+tags: [interview, fulfilment]
+tracks: [okf]
+status: processed
+---
+
+# Summary
+
+The CFO described manual hand-offs between warehouse and finance as the primary bottleneck.

migrations/5.1-to-6.0/fixtures/after/knowledge/order-fulfilment-bottleneck.md

@@ -0,0 +1,19 @@
+---
+type: insight
+title: "Order fulfilment bottleneck"
+description: "Manual hand-offs between warehouse and finance slow order fulfilment."
+source: _intake/processed/2026-07-01-cfo-interview.md
+created_at: 2026-07-01
+confidence: inferred
+tags: [fulfilment, operations]
+superseded_by: /knowledge/warehouse-finance-handoff.md
+timestamp: 2026-07-01T10:00:00Z
+---
+
+# Body
+
+Warehouse picks are complete before finance receives the invoice trigger, creating a multi-day lag.
+
+# Citations
+
+- CFO onboarding interview (`_intake/processed/2026-07-01-cfo-interview.md`)

migrations/5.1-to-6.0/fixtures/after/knowledge/warehouse-finance-handoff.md

@@ -0,0 +1,19 @@
+---
+type: finding
+title: "Warehouse-finance handoff delay"
+description: "Finance invoice trigger arrives days after warehouse completion."
+source: _intake/processed/2026-07-01-cfo-interview.md
+created_at: 2026-07-01
+confidence: observed
+tags: [finance, warehouse]
+supersedes: /knowledge/order-fulfilment-bottleneck.md
+timestamp: 2026-07-01T10:30:00Z
+---
+
+# Body
+
+The hand-off relies on a spreadsheet emailed twice weekly; no system integration exists.
+
+# Citations
+
+- [Order fulfilment bottleneck](/knowledge/order-fulfilment-bottleneck.md)

migrations/5.1-to-6.0/fixtures/after/transitrix.yaml

@@ -0,0 +1,2 @@
+# Upgrade only the methodology pin.
+methodology_version: "6.0.0" # reviewed version

migrations/5.1-to-6.0/fixtures/before/_intake/processed/2026-07-01-cfo-interview.md

@@ -0,0 +1,16 @@
+---
+type: source-document
+title: "CFO onboarding interview"
+description: "First-hand account of the order-fulfilment pain points from the CFO."
+source: "_intake/originals/2026-07-01-cfo-interview.md"
+source_hash: "sha256:0000000000000000000000000000000000000000000000000000000000000001"
+created_at: 2026-07-01
+confidence: observed
+tags: [interview, fulfilment]
+tracks: [okf]
+status: processed
+---
+
+# Summary
+
+The CFO described manual hand-offs between warehouse and finance as the primary bottleneck.

migrations/5.1-to-6.0/fixtures/before/knowledge/order-fulfilment-bottleneck.md

@@ -0,0 +1,19 @@
+---
+type: insight
+title: "Order fulfilment bottleneck"
+description: "Manual hand-offs between warehouse and finance slow order fulfilment."
+source: _intake/processed/2026-07-01-cfo-interview.md
+created_at: 2026-07-01
+confidence: inferred
+tags: [fulfilment, operations]
+superseded_by: /knowledge/warehouse-finance-handoff.md
+timestamp: 2026-07-01T10:00:00Z
+---
+
+# Body
+
+Warehouse picks are complete before finance receives the invoice trigger, creating a multi-day lag.
+
+# Citations
+
+- CFO onboarding interview (`_intake/processed/2026-07-01-cfo-interview.md`)

migrations/5.1-to-6.0/fixtures/before/knowledge/warehouse-finance-handoff.md

@@ -0,0 +1,19 @@
+---
+type: finding
+title: "Warehouse-finance handoff delay"
+description: "Finance invoice trigger arrives days after warehouse completion."
+source: _intake/processed/2026-07-01-cfo-interview.md
+created_at: 2026-07-01
+confidence: observed
+tags: [finance, warehouse]
+supersedes: /knowledge/order-fulfilment-bottleneck.md
+timestamp: 2026-07-01T10:30:00Z
+---
+
+# Body
+
+The hand-off relies on a spreadsheet emailed twice weekly; no system integration exists.
+
+# Citations
+
+- [Order fulfilment bottleneck](/knowledge/order-fulfilment-bottleneck.md)

migrations/5.1-to-6.0/fixtures/before/transitrix.yaml

@@ -0,0 +1,2 @@
+# Upgrade only the methodology pin.
+methodology_version: "5.1.0" # reviewed version