open/ktg-plugin-marketplace
1
0
Fork 0
Code Issues Pull requests Releases Wiki Activity Actions

docs(architect): weekly KB update — 66 files refreshed (2026-04)

Browse source 
Updated 66 stale knowledge base reference files (10 critical, 56 high)
across all 5 skills using Microsoft Learn MCP research.

Key factual updates:
- Groundedness Detection API: `correction` → `mitigating` param,
  `correctedText` → `correctionText` (breaking change)
- Copilot Studio: GPT-4.1 mini now default (was GPT-4o mini);
  Claude Sonnet 4.5 + Opus 4.5 added (experimental, 200K ctx)
- Agentic Retrieval: still public preview; 50M free tokens/month
- Azure security baselines: "Cognitive Services" → "Foundry Tools"
- Databricks: Delta Live Tables → Lakeflow Spark Declarative Pipelines
- MLflow 3 GenAI: new Feedback/Expectation data model
- Token tracking doc: "Azure OpenAI in Foundry Models through a gateway"
- Agent Registry: Risks column (M365 E7), Graph API (preview)
- Copilot DLP: new Entra AI Admin + Purview Data Security AI Admin roles
- ISO/IEC 42001: scope expanded to M365 Copilot, Foundry, Security Copilot
- Zero Trust: CAE now via Conditional Access, Strict Location Enforcement
- Purview: new Fabric Copilots/agents governance section
- AG-UI HITL: ApprovalRequiredAIFunction (C#), @tool approval_mode (Python)

All files: Last updated → 2026-04, *(Verified MCP 2026-04)* markers added.
Build registry: 1341 URLs from 387 files (+2 new URLs).

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This commit is contained in:
Kjell Tore Guttormsen 2026-04-09 22:41:26 +02:00
commit ad8a411f38
73 changed files with 727 additions and 301 deletions
Download patch file Download diff file Expand all files Collapse all files

4
plugins/ms-ai-architect/skills/ms-ai-security/references/cost-optimization/cost-allocation-chargeback.md
View file

@ -1,6 +1,6 @@
# Cost Allocation and Chargeback Models
**Last updated:** 2026-02
**Last updated:** 2026-04
**Status:** GA
**Category:** Cost Optimization & FinOps for AI
@ -394,6 +394,8 @@ Microsoft FinOps Toolkit er **open source** og gratis:
---
*(Verified MCP 2026-04)*
## Kilder og verifisering
### Microsoft Learn (Verified fra MCP Research)

12
plugins/ms-ai-architect/skills/ms-ai-security/references/cost-optimization/gpt5-gpt41-pricing-models.md
View file

@ -1,6 +1,6 @@
# GPT-5 og GPT-4.1: Prismodeller og kostnadsoptimalisering
**Last updated:** 2026-02
**Last updated:** 2026-04
**Status:** GA (GPT-4.1-serien), GA (GPT-5-serien, begrenset tilgang for gpt-5 og gpt-5-codex)
**Category:** Cost Optimization & FinOps for AI
@ -186,14 +186,20 @@ Modeller i Copilot Studio og AI Builder (Power Platform) prises etter tre takstn
| `gpt-5-chat` (preview) | **Standard** | Moderat forbruk | Moderat forbruk |
| `gpt-5-reasoning` (preview) | **Premium** | Høyeste forbruk | Høyeste forbruk |
| `o3` | **Premium** | Høyeste forbruk | Høyeste forbruk |
| `Claude Sonnet 4.5` (experimental) | **Standard** | Moderat forbruk | Moderat forbruk |
| `Claude Opus 4.5` (experimental) | **Premium** | Høyeste forbruk | Høyeste forbruk |
*(Verified MCP 2026-04)*
**Viktige implikasjoner:**
- **GPT-4.1 mini er nå standardmodellen i Copilot Studio** (erstattet GPT-4o mini). GPT-4o og GPT-4o mini er kun tilgjengelig i US government-regioner.
- Copilot Studio inkluderer et månedlig kvantum av Copilot Credits. Å bruke gpt-5-reasoning eller o3 tapper disse vesentlig raskere enn gpt-4.1-mini.
- Standard-rate (gpt-4.1 og gpt-5-chat) er tilgjengelig uten ekstra tilleggslisens i de fleste planer.
- Premium-rate (gpt-5-reasoning, o3) kan kreve pay-as-you-go-overskudd ved høyt volum.
- Premium-rate (gpt-5-reasoning, o3, Claude Opus 4.5) kan kreve pay-as-you-go-overskudd ved høyt volum.
- **Claude Sonnet 4.5 og Opus 4.5** er nå tilgjengelig i Copilot Studio (experimental, 200K kontekstvindu).
- **M365 Copilot (enterprise):** Inkluderer standardtilgang til GPT-5 (inkl. standard Copilot Chat). Priority Access krever M365 Copilot-lisens.
**Confidence:** Høy (basert på offisiell AI Builder/Copilot Studio-dokumentasjon, februar 2026)
**Confidence:** Høy (basert på offisiell AI Builder/Copilot Studio-dokumentasjon, 2026-04)
---

4
plugins/ms-ai-architect/skills/ms-ai-security/references/cost-optimization/inference-endpoint-cost-optimization.md
View file

@ -1,6 +1,6 @@
# Managed Inference Endpoints: Cost Optimization
**Last updated:** 2026-02
**Last updated:** 2026-04
**Status:** GA
**Category:** Cost Optimization & FinOps for AI
@ -558,6 +558,8 @@ Total kostnad = (Instance hours × Instance price)
**ROI-indikator:**
Hvis inference-kostnad per prediction >10% av business value per prediction, er det rom for optimalisering.
*(Verified MCP 2026-04)*
## Kilder og verifisering
### Microsoft Learn-dokumentasjon (Verified via MCP)

14
plugins/ms-ai-architect/skills/ms-ai-security/references/cost-optimization/multi-model-strategy-costs.md
View file

@ -1,6 +1,6 @@
# Multi-Model Strategy: Cost-Performance Trade-offs
**Last updated:** 2026-02
**Last updated:** 2026-04
**Status:** GA
**Category:** Cost Optimization & FinOps for AI
@ -637,11 +637,11 @@ az consumption usage list --start-date 2026-02-01 --end-date 2026-02-28 \
## Kilder og verifisering
**Microsoft Learn (MCP-verified):**
1. [Model router for Azure AI Foundry](https://learn.microsoft.com/en-us/azure/ai-foundry/openai/concepts/model-router) — **Verified** (MCP fetch, 2026-02)
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, 2026-02)
3. [Understanding costs associated with provisioned throughput units (PTU)](https://learn.microsoft.com/en-us/azure/ai-foundry/openai/how-to/provisioned-throughput-onboarding) — **Verified** (MCP search, 2026-02)
4. [Azure OpenAI in Azure AI Foundry Models](https://learn.microsoft.com/en-us/azure/ai-foundry/openai/concepts/models) — **Verified** (MCP search, 2026-02)
5. [GPT-4o vs GPT-4o mini model selection](https://learn.microsoft.com/en-us/azure/ai-foundry/openai/whats-new) — **Verified** (MCP search, 2026-02)
1. [Model router for Azure AI Foundry](https://learn.microsoft.com/en-us/azure/ai-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)
3. [Understanding costs associated with provisioned throughput units (PTU)](https://learn.microsoft.com/en-us/azure/ai-foundry/openai/how-to/provisioned-throughput-onboarding) — **Verified** (MCP search, 2026-04)
4. [Azure OpenAI in Azure AI Foundry Models](https://learn.microsoft.com/en-us/azure/ai-foundry/openai/concepts/models) — **Verified** (MCP search, 2026-04)
5. [GPT-4o vs GPT-4o mini model selection](https://learn.microsoft.com/en-us/azure/ai-foundry/openai/whats-new) — **Verified** (MCP search, 2026-04)
**GitHub samples (MCP-referenced):**
1. [Smart load balancing for Azure OpenAI (Azure API Management)](https://github.com/Azure-Samples/openai-apim-lb) — **Verified**
@ -666,6 +666,6 @@ az consumption usage list --start-date 2026-02-01 --end-date 2026-02-28 \
---
**Sist oppdatert:** 2026-02 (basert på Model Router version `2025-11-18` og Azure OpenAI pricing per februar 2026)
**Sist oppdatert:** 2026-04 (basert på Model Router version `2025-11-18` og Azure OpenAI pricing per april 2026). Verified (MCP 2026-04).
**Neste review:** Ved nye Model Router-versjoner eller større pricing changes.

16
plugins/ms-ai-architect/skills/ms-ai-security/references/cost-optimization/observability-cost-reduction.md
View file

@ -1,6 +1,6 @@
# Observability and Monitoring Cost Optimization
**Last updated:** 2026-02
**Last updated:** 2026-04
**Status:** GA
**Category:** Cost Optimization & FinOps for AI
@ -39,7 +39,7 @@ Moderne Azure Monitor tilbyr flere kostnadseffektive alternativer som Basic Logs
| Strategi | Mekanisme | Bruksområde | Trade-off |
|----------|-----------|-------------|-----------|
| **Adaptive Sampling** | Automatisk justering basert på telemetri-volum (default: 5 items/sec) | ASP.NET, ASP.NET Core, Azure Functions | Reduserer volum uten konfigurasjon, kan miste sjeldne events |
| **Adaptive Sampling** | Automatisk justering basert på telemetri-volum (default: 5 items/sec) — gjelder **kun Classic API SDK** (ASP.NET, ASP.NET Core). OpenTelemetry-baserte distros har **ikke** adaptive sampling aktivert som default. | ASP.NET, ASP.NET Core, Azure Functions | Reduserer volum uten konfigurasjon (Classic SDK kun), kan miste sjeldne events | Verified (MCP 2026-04) |
| **Fixed-rate Sampling** | Fast prosentandel (f.eks. 10%, 25%, 50%) | Konsistent sampling på tvers av client/server | Forutsigbar reduksjon, krever manuell tuning |
| **Rate-limited Sampling** | Begrenser til maks N requests/sec (f.eks. 1.5 req/sec) | Java-applikasjoner, cost-capping | Streng volum-kontroll, kan miste spikes |
| **Ingestion Sampling** | Server-side sampling (kun hvis SDK ikke sampler) | Legacy apps uten SDK-sampling | Reduserer ikke nettverkstrafikk |
@ -95,7 +95,7 @@ builder.Services.AddApplicationInsightsTelemetry(new ApplicationInsightsServiceO
**Kostnad:** 50-70% reduksjon vs. full observability.
**Eksempel (Java Agent 3.4+, rate-limited sampling):**
**Eksempel (Java Agent 3.7.5+, rate-limited sampling):**
```json
{
"sampling": {
@ -401,7 +401,7 @@ For volumer >1 TB/dag, vurder dedicated cluster for ytterligere besparelser (clu
## Kilder og verifisering
### Microsoft Learn-dokumentasjon (Verified via MCP 2026-02)
### Microsoft Learn-dokumentasjon (Verified via MCP 2026-04)
1. **Sampling in Application Insights:**
https://learn.microsoft.com/en-us/azure/azure-monitor/app/sampling-classic-api
@ -455,10 +455,10 @@ For volumer >1 TB/dag, vurder dedicated cluster for ytterligere besparelser (clu
| Seksjon | Confidence | Kilde |
|---------|------------|-------|
| Sampling-strategier | **Verified** | Microsoft Learn MCP (feb 2026) |
| Prismodell | **Verified** | Microsoft Learn MCP (feb 2026) |
| Table plans | **Verified** | Microsoft Learn MCP (feb 2026) |
| Retention policies | **Verified** | Microsoft Learn MCP (feb 2026) |
| Sampling-strategier | **Verified** | Microsoft Learn MCP (apr 2026) |
| Prismodell | **Verified** | Microsoft Learn MCP (apr 2026) |
| Table plans | **Verified** | Microsoft Learn MCP (apr 2026) |
| Retention policies | **Verified** | Microsoft Learn MCP (apr 2026) |
| Arkitektuurmønstre | **Baseline** | Kombinasjon av verified docs + modellkunnskap |
| Norsk compliance | **Baseline** | Lovtekst + modellkunnskap (krever juridisk validering) |
| Kostnadseksempler (NOK) | **Baseline** | Estimater basert på Azure pricing calculator (feb 2026) |

9
plugins/ms-ai-architect/skills/ms-ai-security/references/cost-optimization/rag-query-cost-reduction.md
View file

@ -1,6 +1,6 @@
# RAG Query Cost Optimization
**Last updated:** 2026-02
**Last updated:** 2026-04
**Status:** GA
**Category:** Cost Optimization & FinOps for AI
@ -142,8 +142,14 @@ else: # Complex reasoning
### 4. Agentic Retrieval (Cost-Aware)
**Status: Public Preview** — Agentic Retrieval er foreløpig i public preview (ikke GA). *(Verified MCP 2026-04)*
**Prinsipp:** Azure AI Search Agentic Retrieval bruker LLM til å generere subqueries som kjøres parallelt. Dette kan være dyrt, men også mer effektivt enn multiple sequential queries.
**Prismodell (public preview):**
- **Free tier:** 50 millioner gratis agentic reasoning tokens/måned inkludert (på Basic tier og høyere)
- **Standard tier:** Pay-as-you-go etter at gratis kvota er brukt
**Kostnadseksempel (Verified - Microsoft Learn):**
- **2000 agentic retrievals** med 3 subqueries per plan:
- Reranking: ~$3.30 (150M tokens @ $0.022/token)
@ -156,6 +162,7 @@ else: # Complex reasoning
- Når answer quality er kritisk og kostnaden kan rettferdiggjøres.
**Cost control:**
- 50M gratis tokens/mnd dekker typisk moderat bruk i utviklings- og testmiljøer.
- Sett `reasoning_effort` til `minimal` eller `low` (ikke `medium`).
- Begrens antall subqueries per plan.

6
plugins/ms-ai-architect/skills/ms-ai-security/references/cost-optimization/reserved-capacity-planning.md
View file

@ -1,6 +1,6 @@
# Reserved Capacity and Commitment Discounts
**Last updated:** 2026-02
**Last updated:** 2026-04
**Status:** GA
**Category:** Cost Optimization & FinOps for AI
@ -21,7 +21,7 @@ Reserved capacity og commitment tier pricing er Azures to primære mekanismer fo
| **Deployment types** | Regional, Data Zone, Global Provisioned | Web API, Connected containers, Disconnected containers |
| **Overage håndtering** | Hourly rate for excess PTUs | Overage rate per commitment tier |
**💡 Confidence: HIGH** — Basert på offisiell Microsoft dokumentasjon oppdatert januar 2026.
**💡 Confidence: HIGH** — Basert på offisiell Microsoft dokumentasjon oppdatert april 2026. Verified (MCP 2026-04).
---
@ -279,6 +279,8 @@ Er tjenesten Azure OpenAI/Foundry Models?
**💡 Best Practice:** **ALLTID deploy først, kjøp reservasjon etterpå.** Quota ≠ capacity.
**Autorenew er nå ON som standard for alle nye reservasjoner** (gjelder reservasjoner opprettet etter 2025-Q4). Sjekk innstillingen ved kjøp og deaktiver manuelt hvis ønskelig. Verified (MCP 2026-04).
---
### Sizing og kapasitetsplanlegging

4
plugins/ms-ai-architect/skills/ms-ai-security/references/cost-optimization/small-language-models-economics.md
View file

@ -1,6 +1,6 @@
# Small Language Models: Economics and Use Cases
**Last updated:** 2026-02
**Last updated:** 2026-04
**Status:** GA
**Category:** Cost Optimization & FinOps for AI
@ -576,6 +576,8 @@ az webapp create --name webapp-slm-phi4 --resource-group rg-slm-norway --plan pl
| **Team: Begynner** | ✅ | ✅ | ❌ | ❌ |
| **Team: Ekspert** | ✅ | ✅ | ✅ | ✅ |
*(Verified MCP 2026-04)*
## Kilder og verifisering
### Microsoft Learn (MCP-verified, 2026-02)

4
plugins/ms-ai-architect/skills/ms-ai-security/references/cost-optimization/vector-storage-cost-optimization.md
View file

@ -1,6 +1,6 @@
# Vector Storage and Embedding Cost Optimization
**Last updated:** 2026-02
**Last updated:** 2026-04
**Status:** GA
**Category:** Cost Optimization & FinOps for AI
@ -522,6 +522,8 @@ Hvis vector search brukes som grunnlag for Copilot for Microsoft 365:
- Fine-tuned embedding-modeller for domenet
- **Mål:** Maksimal ROI, konkurransefortrinn
*(Verified MCP 2026-04)*
## Kilder og verifisering
### Microsoft Learn (MCP-verifisert, februar 2026)