open/ms-ai-architect
1
0
Fork 0
Code Issues Releases 1 Activity

feat(ms-ai-architect): anvend form #36 paa alle 5 medlemmene; fell 663(c) som #37-medlem 6 [skip-docs]

Browse source 
Form #36 (droppet kvalifikator under et verifikasjonsstempel) anvendt paa sine
fem maalte medlemmer. Reparasjonen er en INNSNEVRING i hvert tilfelle, aldri
rutet gjennom #34/#35. Fersk microsoft-learn MCP-henting per medlem, gjort av
HOVEDKONTEKSTEN (ingen subagenter). Hel-linje-unikhet talt korpus-bredt (394
filer) foer hver erstatning; alle fem = 1 forekomst.

Korpus-bred kildemaaling (#35's regel), nevner rapportert: 10 linjer i 9 filer
siterer azure-openai-gateway-multi-backend. Av de 8 oevrige baerer INGEN en
paastand om credential termination eller topologi-omfang - defekten er begrenset
til medlem 2 og 3.

1. model-versioning-registry-management.md:323 - "For production online endpoint
   deployments" -> "For production workloads paa managed online endpoints".
   idx-33t lukket (status=resolved + resolution).
2. ai-services-enterprise-architecture.md:573 - BEGGE former i EN edit, ett
   stempelflytt: #36-leddet innsnevret til topologi (1); #37-leddet (tagg-
   paastanden) fjernet.
3. multi-model-strategy-costs.md:663 - #36 paa (a)+(b), "Dokument bekrefter" ->
   "Dokumentet bekrefter, for topologien ...". Ledd (c) ADJUDISERT OG FELT i
   samme henting: sida har 0 treff paa "Foundry Tools", 0 paa "Azure OpenAI in
   Foundry Models" og 0 paa "Azure OpenAI" i det hele tatt, mot kjent-positiv-
   kontroll "credential termination"=1 / "Foundry"=11. Foert inn i #37's
   GOVERNS som sjette medlem, ren append, datert.
4. genaiops-llm-specific-practices.md:212 - "(frittstaaende, anbefalt)" ->
   "(frittstaaende)" + kildens egen kategori-preferanse mot egenutviklet kode.
   Blokkstempelets naboer (211, 213) verifisert mot samme henting foer flytt.
5. multimodal-prompt-design.md:193 - "AI-genererte bildebeskrivelser" ->
   samme med kildens vilkaar (modelName/modelDeployment, 2026-05-01-preview).
   Blokkstempelet leste allerede 2026-08-25 = hentedato, saa det ble IKKE
   flyttet - ingen ny paastand avgitt om linje 196 (idx-33v's aapne medlem).

NYTT, UMAALT LEDD FUNNET UNDER ANVENDELSE: chargeback/usage-tracking-leddet paa
573 og 663(b) baerer samme #36-defekt - "Using a gateway in this topology
enables client-based usage tracking" staar under topologi 1, og sida sier det
motsatte for topologi 2. Stempelet som flyttes dekker det leddet, saa det ble
innsnevret i samme edit. Dette falsifiserer #37's egen medlem-2-frikjennelse
("credential termination and the chargeback point ARE faithful"); korreksjonen
er foert i #37's GOVERNS-tillegg, siden ordren forbyr aa redigere formteksten.

Gater: check-g7-queue.mjs 0 drift exit 0 - suite 1052/1052.
Assertert i skriptet: alle 79 oevrige entries bit-identiske, idx-33t endret kun
status+resolution, _meta urort utenfor ratified_forms["#37"], #37 ren append.
This commit is contained in:
Kjell Tore Guttormsen 2026-08-25 23:40:20 +02:00
commit 147ca0b993
6 changed files with 10 additions and 9 deletions
Download patch file Download diff file Expand all files Collapse all files

7
scripts/kb-eval/data/g7-review-queue.json
View file

File diff suppressed because one or more lines are too long

2
skills/ms-ai-advisor/references/prompt-engineering/multimodal-prompt-design.md
View file

@ -190,7 +190,7 @@ messages = [
**Azure AI Search multimodal pipeline (Verified MCP 2026-08-25):**
1. **Content extraction** — to anbefalte innebygde skills:
- Document Extraction skill: rask prototyping/produksjon der eksakt posisjon ikke kreves; bilde-posisjonsmetadata kun for PDF; ingen innebygd chunking (bruk Text Split skill)
- Azure Content Understanding skill: avansert — cross-page tabeller, semantisk chunking (innebygd), AI-genererte bildebeskrivelser, og tekst-/bilde-posisjonsmetadata for blant annet PDF, DOCX, XLSX og PPTX
- Azure Content Understanding skill: avansert — cross-page tabeller, semantisk chunking (innebygd), AI-genererte bildebeskrivelser (når `modelName` og `modelDeployment` er konfigurert, tilgjengelig fra REST API `2026-05-01-preview`), og tekst-/bilde-posisjonsmetadata for blant annet PDF, DOCX, XLSX og PPTX
- (Document Layout skill er fortsatt støttet for *eksisterende* pipelines, men for nye skillsets anbefaler Microsoft Azure Content Understanding skill, som slår sammen ekstraksjon og chunking i én skill.)
2. **Text chunking:** Text Split skill (ikke nødvendig med Content Understanding, som chunker semantisk)
3. **Image verbalization:** GenAI Prompt skill + LLM (phi-4, gpt-4o, gpt-5) → naturlig-språklig beskrivelse

2
skills/ms-ai-engineering/references/azure-ai-services/ai-services-enterprise-architecture.md
View file

@ -570,7 +570,7 @@ TOTAL: ~46 700 NOK/måned (høyere cost, men forutsigbar)
3. [Baseline Foundry Chat Architecture (Foundry Agent Service + Microsoft Agent Framework)](https://learn.microsoft.com/en-us/azure/architecture/ai-ml/architecture/baseline-microsoft-foundry-chat) — Verified (MCP 2026-04)
4. [Azure API Management - AI Gateway Capabilities](https://learn.microsoft.com/en-us/azure/api-management/genai-gateway-capabilities)
5. [Reliability in Azure AI Search](https://learn.microsoft.com/en-us/azure/reliability/reliability-ai-search)
6. [Multi-Backend Gateway Guide](https://learn.microsoft.com/en-us/azure/architecture/ai-ml/guide/azure-openai-gateway-multi-backend) — Verified MCP 2026-04: Dokumentet bekrefter fire gateway-topologier: (1) multiple model deployments i single instance, (2) multiple instances same region/single subscription, (3) multiple instances same region/multiple subscriptions (eksplisitt som egen topologi), (4) multiple instances multi-region. Tagger eksplisitt "Foundry Tools" og "Azure OpenAI in Foundry Models". Anbefaler sterkt credential termination og reestablishment ved gateway fremfor pass-through client credentials. Gateway muliggjør client-based usage tracking for chargeback-modeller. Verified (MCP 2026-04)
6. [Multi-Backend Gateway Guide](https://learn.microsoft.com/en-us/azure/architecture/ai-ml/guide/azure-openai-gateway-multi-backend) — Verified MCP 2026-08-25: Dokumentet bekrefter fire gateway-topologier: (1) multiple model deployments i single instance, (2) multiple instances same region/single subscription, (3) multiple instances same region/multiple subscriptions (eksplisitt som egen topologi), (4) multiple instances multi-region. For topologi (1), «multiple model deployments in a single instance», anbefaler dokumentet sterkt credential termination og reestablishment ved gateway fremfor pass-through client credentials, og det er for samme topologi det sier at gateway gir client-based usage tracking for chargeback-modeller. Verified (MCP 2026-08-25)
7. [Load Balancing Options - Azure Architecture](https://learn.microsoft.com/en-us/azure/architecture/guide/technology-choices/load-balancing-overview)
**GitHub Samples (Microsoft-verified):**

4
skills/ms-ai-engineering/references/mlops-genaiops/genaiops-llm-specific-practices.md
View file

@ -207,9 +207,9 @@ MLflow Tracing provides end-to-end observability for GenAI applications:
- **Federated authentication**: Extend client auth beyond Entra ID and API keys
- **Data sovereignty**: Enforce regional routing compliance for GDPR
**Implementasjonsalternativer (Verified MCP 2026-08):**
**Implementasjonsalternativer (Verified MCP 2026-08-25):**
1. **Foundry AI Gateway (innebygd)** — konfigureres i Foundry-portalen, bygger på Azure API Management. Støtter ingress til ÉN Foundry-ressurs og spenner ikke over flere ressurser.
2. **Azure API Management** (frittstående, anbefalt) — PaaS, built-in AI-policies (`llm-token-limit`, `llm-emit-token-metric`, `llm-content-safety`, `llm-semantic-cache-store`), zone-redundant, multi-region. Bruk [GenAI gateway toolkit](https://github.com/Azure-Samples/apim-genai-gateway-toolkit) for custom policies + load-testing.
2. **Azure API Management** (frittstående) — PaaS, built-in AI-policies (`llm-token-limit`, `llm-emit-token-metric`, `llm-content-safety`, `llm-semantic-cache-store`), zone-redundant, multi-region. API Management er generelt det foretrukne alternativet fremfor egenutviklet kode, enten integrert med Foundry eller frittstående. Bruk [GenAI gateway toolkit](https://github.com/Azure-Samples/apim-genai-gateway-toolkit) for custom policies + load-testing.
3. **Custom code** — Deploy gateway-logikk til App Service, Container Apps eller AKS. Kan frontes av API Management for HTTP-gateway capabilities.
**Viktig:** Global og data zone deployments i Azure OpenAI (som distribuerer kapasitet på tvers av datasentre) er i seg selv en gateway-implementasjon — vurder om disse dekker behovet FØR du legger til ekstra gateway-lag.

2
skills/ms-ai-engineering/references/mlops-genaiops/model-versioning-registry-management.md
View file

@ -320,7 +320,7 @@ Microsoft Foundry Model Catalog bruker samme underliggende registry-infrastruktu
**Scenario:** Registrer Custom AI Builder model i Azure ML Registry for reuse.
> **Merk (Verified MCP 2026-04):** For production online endpoint deployments anbefaler Microsoft nå Microsoft Entra token-based authentication (`aad_token`) fremfor key-based authentication for forbedret sikkerhet via identity-based access control.
> **Merk (Verified MCP 2026-08-25):** For production workloads på managed online endpoints anbefaler Microsoft nå Microsoft Entra token-based authentication (`aad_token`) fremfor key-based authentication for forbedret sikkerhet via identity-based access control.
- Tren modell i AI Builder
- Eksporter modell (hvis tilgjengelig)
- Registrer i Azure ML Registry som MLflow model

2
skills/ms-ai-security/references/cost-optimization/multi-model-strategy-costs.md
View file

@ -660,7 +660,7 @@ az consumption usage list --start-date 2026-02-01 --end-date 2026-02-28 \
**Microsoft Learn (MCP-verified):**
1. [Model router for Microsoft Foundry](https://learn.microsoft.com/en-us/azure/foundry/openai/concepts/model-router) — **Verified** (MCP fetch, 2026-04)
2. [Use a gateway in front of multiple Azure OpenAI deployments](https://learn.microsoft.com/en-us/azure/architecture/ai-ml/guide/azure-openai-gateway-multi-backend) — **Verified** (MCP fetch, 2026-04). Dokument bekrefter: (a) credential termination og reestablishment ved gateway anbefales fremfor pass-through client credentials, (b) gateway gir client-based usage tracking og chargeback-støtte, (c) Azure OpenAI er nå tagget som "Foundry Tools / Azure OpenAI in Foundry Models".
2. [Use a gateway in front of multiple Azure OpenAI deployments](https://learn.microsoft.com/en-us/azure/architecture/ai-ml/guide/azure-openai-gateway-multi-backend) — **Verified** (MCP fetch, 2026-08-25). Dokumentet bekrefter, for topologien «multiple model deployments in a single instance»: (a) credential termination og reestablishment ved gateway anbefales sterkt fremfor pass-through client credentials, (b) gateway gir client-based usage tracking og chargeback-støtte.
3. [Understanding costs associated with provisioned throughput units (PTU)](https://learn.microsoft.com/en-us/azure/foundry/openai/concepts/provisioned-throughput-billing) — **Verified** (MCP search, 2026-04)
4. [Azure OpenAI in Foundry Models](https://learn.microsoft.com/en-us/azure/foundry/foundry-models/concepts/models-sold-directly-by-azure) — **Verified** (MCP search, 2026-04)
5. [GPT-4o vs GPT-4o mini model selection](https://learn.microsoft.com/en-us/azure/foundry-classic/openai/whats-new) — **Verified** (MCP search, 2026-04)