fix(ms-ai-architect): idx-26ao var blokkform hele tiden, og en sjette enumerasjon sto utenfor de 58

To funn fra arbeidet med beslutningsgrunnlaget for #21-#24, bokfoert foer
noen beslutning tas. Ingen korpus-edit.

idx-26ao: datasettet klassifiserte fila som line-form. Den er blokkform -
"### MCP-kall utfoert" over tre verktoey-bullets (3+3+2=8) rett over
mikslinja, samme form som idx-26z, som SAMME datasett klassifiserte som
block. De tre bulletene var ikke ankret. Aa anvende «slett linja» paa
:383 alene ville latt overskriften staa over tre kall-bullets - idx-26aq-
residuet paa nytt, denne gang produsert av en entry som ikke ankret sine
egne overlevende. Fire ankere lagt til; entryen hoerer under #22 i tillegg
til #23. Funnet ved aa sveipe alle 46 line-form-medlemmer for en skjult
enumerasjon over ankeret: 3 treff, 2 nett-stoey, 1 ekte. Én av 46, saa
feilen er isolert - maalt, ikke antatt.

idx-26ar (ny): zero-etl-fabric-patterns.md baerer kall-enumerasjonen uten
noe tall, og var derfor aldri i de 58 - klassen er definert som en stated
COUNT. Den falsifiserer idx-26an sin sizing («5 av 389 ... formen er naa
fullt enumerert») med én. Nettet her krevde et listeelement som navngir et
verktoey, ikke et siffer ved siden av det; 15 filer, alle haandlest, 12
allerede bokfoert, 2 falske positive, denne ubokfoert. Tre former i én fil:
kall-enumerasjonen (:629-632), en innholdsinventar-etikett attributtert til
et verktoey (:624), og en provenance-overclaim i keep-klassen (:618, kilde 4
er en azure.microsoft.com-prisside under en microsoft-learn-etikett).

Maalt underveis, og det avgjoer #23: keep-halvdelen i idx-26ao er SANN og
gjentatt to steder til (18 nummererte kilder, :6 og :356), mens den i
idx-26z er FALSK (3 paastaatt mot 5 rader i filas egen Verified-tabell).
I begge tilfeller holder den ratifiserte «slett linja» uten aa forfatte noe.

Suite 1052/1052. Alle 52 entries ankrer til live korpus.

scripts/kb-eval/data/g7-review-queue.json

@@ -824,12 +824,16 @@
       "class": "multi-locator",
       "status": "open",
       "raised": "2026-08-11",
-      "summary": "PULLED OUT OF THE CLASS ENTRY idx-26an the same session it was written. It was classified 2026-08-11 as line-form with no flag and no open decision, and a residue screen run before any corpus edit falsified that. MEASURED at :383: \"**Totalt:** 18 kilder, 8 MCP-kall.\" THE LINE IS MIXED. 8 MCP-kall is the delete class - an unverifiable call count. 18 kilder is the KEEP class, and unlike the call count it is CHECKABLE: the file carries a heading \"### Microsoft Learn-kilder (18 dokumenter)\" at :356 over a numbered source list. So \"delete the line\", the ratified form, would delete a verifiable claim along with an unverifiable one. This is the same structural defect as idx-26z, where \"**Totalt:** 4 MCP-kall, 3 unike Microsoft Learn-kilder.\" mixes the two classes in one sentence - and the same hazard idx-26v named: splitting the line means AUTHORING a replacement, which is how a repair authors the defect class it closes. THE RATIFIED FORM (idx-26j, 2026-08-09) still governs the call count; what is undecided is the scope of the edit on a mixed line. Note also :6, \"**Konfidensgrad:** Høy (basert på 18 MCP-kilder fra Microsoft Learn)\" - the header restates the 18. That is the KEEP-class number, so it is not a residue of the call count, but any re-authoring of :383 has to leave :6 coherent with it. Decide together with idx-26z.",
+      "summary": "PULLED OUT OF THE CLASS ENTRY idx-26an the same session it was written. It was classified 2026-08-11 as line-form with no flag and no open decision, and a residue screen run before any corpus edit falsified that. MEASURED at :383: \"**Totalt:** 18 kilder, 8 MCP-kall.\" THE LINE IS MIXED. 8 MCP-kall is the delete class - an unverifiable call count. 18 kilder is the KEEP class, and unlike the call count it is CHECKABLE: the file carries a heading \"### Microsoft Learn-kilder (18 dokumenter)\" at :356 over a numbered source list. So \"delete the line\", the ratified form, would delete a verifiable claim along with an unverifiable one. This is the same structural defect as idx-26z, where \"**Totalt:** 4 MCP-kall, 3 unike Microsoft Learn-kilder.\" mixes the two classes in one sentence - and the same hazard idx-26v named: splitting the line means AUTHORING a replacement, which is how a repair authors the defect class it closes. THE RATIFIED FORM (idx-26j, 2026-08-09) still governs the call count; what is undecided is the scope of the edit on a mixed line. Note also :6, \"**Konfidensgrad:** Høy (basert på 18 MCP-kilder fra Microsoft Learn)\" - the header restates the 18. That is the KEEP-class number, so it is not a residue of the call count, but any re-authoring of :383 has to leave :6 coherent with it. Decide together with idx-26z.\nADDENDUM 2026-08-11, later the same day - a dated correction, not a silent reopening (sec. 9.2). THE FORM CLASSIFICATION IS WRONG AND THE ANCHOR SET WAS INCOMPLETE. The dataset records this file as form=line. It is BLOCK FORM. MEASURED at :377-383: the heading \"### MCP-kall utfoert\" (spelled with the o-slash in the file) stands over three tool bullets - docs_search 3 soek, docs_fetch 3 hentinger, code_sample_search 2 soek - which sum to 8 and agree with the stated total. That is the same shape as idx-26z (chain-of-thought-prompting.md), which the SAME dataset classified as block. Identical structure, opposite classification.\nCONSEQUENCE, WHICH IS WHY THIS IS BOOKED RATHER THAN NOTED: applying the ratified \"delete the line\" to :383 alone would leave the heading standing over three call bullets asserting the same unverifiable claim - exactly the idx-26aq residue, this time produced by an edit whose own entry failed to anchor the survivors. The four locators are anchored here rather than described in prose, so none survives only in a summary (the 795d494 failure). THIS ENTRY THEREFORE FALLS UNDER OPEN QUESTION #22 AS WELL AS #23, and #22 must be ratified before any deletion is applied to it.\nHOW IT WAS FOUND, and what the net cost: a sweep of ALL 46 line-form members of the 58 for an enumeration hidden above the anchor (a heading naming MCP/calls/research, or two or more tool-naming list items, within 12 lines above). Three hits, hand-read. Two are net noise - inferencing-optimization-caching.md, where the matched heading is a code-samples section and the anchor already carries its own full breakdown on one line, and ai-impact-assessment-framework.md, where the matched lines carry no per-tool counts. This one is real. One of 46, so the misclassification is isolated, not systemic - measured rather than assumed.\nTHE KEEP-CLASS HALF IS MEASURED, NOT ASSUMED, and it settles the #23 question for this file without authoring anything: \"18 kilder\" is TRUE - the file enumerates 18 numbered sources - and it is restated TWICE more, at :6 in the header block and at :356 in the heading \"### Microsoft Learn-kilder (18 dokumenter)\". Deleting the whole mixed line therefore loses no checkable claim. Contrast idx-26z, where the same measurement falsifies the keep half instead: \"3 unike Microsoft Learn-kilder\" against 5 rows in that file's own Verified table.",
       "evidence": "skills/ms-ai-engineering/references/mlops-genaiops/genaiops-llm-specific-practices.md:383 (the mixed total), :356 (the 18-source heading and numbered list), :6 (header restating 18); dataset r11-footer-class-2026-08-11.json (classified form=line, flags=[] - the gap this entry records); idx-26z (same mix), idx-26v (replacement-authors-the-defect hazard)",
       "anchors": [
         "**Totalt:** 18 kilder, 8 MCP-kall.",
         "### Microsoft Learn-kilder (18 dokumenter)",
-        "**Konfidensgrad:** Høy (basert på 18 MCP-kilder fra Microsoft Learn)"
+        "**Konfidensgrad:** Høy (basert på 18 MCP-kilder fra Microsoft Learn)",
+        "### MCP-kall utført",
+        "- **microsoft_docs_search**: 3 søk (GenAIOps overview, LLMOps best practices, lifecycle)",
+        "- **microsoft_docs_fetch**: 3 hentinger (maturity model, genaiops-for-mlops, databricks llmops)",
+        "- **microsoft_code_sample_search**: 2 søk (evaluation Python code, monitoring code)"
       ]
     },
     {
@@ -860,6 +864,23 @@
         "- microsoft_docs_search: \"Azure OpenAI provisioned throughput PTU cost optimization\"",
         "**Unique URLs:** 10"
       ]
+    },
+    {
+      "id": "idx-26ar",
+      "file": "skills/ms-ai-engineering/references/data-engineering/zero-etl-fabric-patterns.md",
+      "class": "multi-locator",
+      "status": "open",
+      "raised": "2026-08-11",
+      "summary": "SURFACED 2026-08-11 while assembling the operator decision basis for open questions #21-#24, from a corpus-wide completeness count of the call-enumeration form. NOT A MEMBER OF THE 58, and the reason is structural: the footer class is defined as \"a stated COUNT of MCP tool calls made while researching THIS document\", and this file states NO count anywhere. It enumerates the calls without totalling them. That definition is why it was never measured into r11-footer-class-2026-08-11.json and has never had a queue entry.\nIT FALSIFIES A SIZING CLAIM, recorded here rather than left standing. idx-26an's residue sizing states that exactly 5 files of 389 carry an enumerated MCP-call list, and concludes \"the form is small and now fully enumerated, not an open-ended sweep\". This is a sixth. THE NET AND ITS COST ARE STATED so the next sweep can be blind somewhere else instead: a list item (leading \"- \" or \"N.\") naming a microsoft-learn / tavily / WebSearch / ToolSearch tool, two or more within three lines, over all 389 files. It returned 15 files and EVERY hit was hand-read. Twelve were already booked or already in the 58. Two are false positives - source-traceability-assumption-register.md:237-238, instructional prose telling the reader how to use MCP tools rather than reporting calls made, and copilot-studio-nlp-configuration.md:553-555, a description of verification method with no per-tool counts. This file is the only unbooked real member. The earlier net required a DIGIT adjacent to the tool name; this file's enumeration carries no digits at all, and that is precisely the blindness that hid it.\nTHREE SHAPES IN ONE FILE, and they are NOT one decision.\n(a) THE CALL ENUMERATION, :629-632. \"**Soek brukt:**\" over three bullets naming docs_search with three query strings, docs_fetch with two targets, and code_sample_search with one query. Under the ratified ground (idx-26j: UNVERIFIABILITY, re-verified independently 2026-08-11 rather than inherited - the only artifacts under scripts/, docs/, tests/ and hooks/ that mention MCP calls are this queue, the footer dataset, docs/r11-pilot-results.md and docs/ref-kb-direction-note-2026-06.md, all analysis and none a run record) this is the same delete class as a bare total, and idx-26aq(a) argues the ground applies with MORE force to a list asserting specific queries than to a number. NOT RATIFIED FOR THIS FILE: whether the enumeration itself is delete class is exactly what open question #22 settles, and #22 was scoped over the 58, which this file is outside.\n(b) THE CODE-SAMPLES LABEL, :624. \"**Kodeeksempler (MCP: microsoft_code_sample_search):**\" attributes a three-item content inventory to a tool. Neither a call count nor a source count - the third category idx-26x named and the ratified form has never addressed.\n(c) A PROVENANCE OVERCLAIM IN THE KEEP CLASS, :618. \"**Primaerkilder (MCP: microsoft-learn):**\" stands over four numbered sources, of which source 4 is https://azure.microsoft.com/pricing/details/microsoft-fabric/ - not learn.microsoft.com, and so not servable by the microsoft-learn MCP server, which serves learn.microsoft.com and nothing else. This is the idx-26w class in a second file and belongs with open question #21. Measured: 4 numbered sources, 6 distinct URLs in the file, 4 of them on learn.microsoft.com.\nNOT REPAIRED. Booked so the file cannot fall out of the programme unnoticed. Every locator is anchored verbatim, and each anchor was verified to occur exactly once AND to equal its whole line (grep-count plus a whole-line comparison, since the gate proves presence and never uniqueness) before this entry was written.",
+      "evidence": "skills/ms-ai-engineering/references/data-engineering/zero-etl-fabric-patterns.md:618-634 (file is 662 lines); absent from scripts/kb-eval/data/r11-footer-class-2026-08-11.json (all 58 members checked by path); idx-26an resolution (the \"5 of 389, fully enumerated\" sizing this falsifies); idx-26aq (the enumeration-is-delete-class argument, and the heading-borne shape); idx-26w (provenance class); idx-26x (content-inventory category); idx-26j resolution and its 2026-08-09 scope addendum (the unverifiability ground)",
+      "anchors": [
+        "**Primærkilder (MCP: microsoft-learn):**",
+        "**Kodeeksempler (MCP: microsoft_code_sample_search):**",
+        "**Søk brukt:**",
+        "- `microsoft_docs_search`: \"Fabric mirroring\", \"zero ETL Fabric\", \"database mirroring OneLake\"",
+        "- `microsoft_docs_fetch`: Mirroring overview, PostgreSQL architecture",
+        "- `microsoft_code_sample_search`: \"Fabric mirroring OneLake Delta Lake\""
+      ]
     }
   ]
 }