open/ms-ai-architect
1
0
Fork 0
Code Issues Releases Activity
244 commits 1 branch 8 tags 135 MiB
Commit graph

5 commits

Author SHA1 Message Date
Kjell Tore Guttormsen
692b3f75d8 docs(ms-ai-architect): KB-refresh critical cost 6/8 — token-counting verifisert mot fine-tuning-cost (RFT) + prompt-caching 
Verifisert mot live Microsoft Learn (foundry/openai/how-to/fine-tuning-cost-management,
prompt-caching, use-your-data):
- RFT (Reinforcement Fine-Tuning) lagt til (filen dekket kun SFT/DPO): tidsbasert «pris = treningstid
  × timepris + grader-tokens», ~$100/time core training for o4-mini-2025-04-16, per-job-tak $5000
  (pauses + deploybart checkpoint ved tak, kan resume uten videre pristak). Cost-control-strategier med.
- Cache-TTL-tabell hadde FEIL akse («Foundry Classic vs AI Services») → korrigert til in-memory
  (5-10 min/maks 1t, alle GPT-4o+) vs extended (24h, GPT-5-serien+gpt-4.1, default for nyere).
- prompt_cache_key ~15 req/min overflow-caveat lagt til.
- On Your Data deprecation 2026-10-14 flagget (3 steder: token-est-seksjon, Foundry-seksjon, kilde #6).
- tiktoken model-set manglet o1-mini/o3-mini → lagt til.
- Header 2026-02→2026-06, verification-date 2026-02-04→2026-06. Priser urørt (kanonisk peker). validate 239/0.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
 2026-06-23 22:53:35 +02:00
Kjell Tore Guttormsen
03d596e4ec docs(ms-ai-architect): KB-refresh tema-b — Foundry-navnesveip «Azure AI Foundry»→«Microsoft Foundry» (233 filer) 
Verifisert mot offisiell MS-doc (juni 2026): «Microsoft Foundry» er det
gjeldende produkt-/portalnavnet; «Foundry (classic)» = gamle «Azure AI Foundry»
(/azure/foundry/ vs /azure/foundry-classic/). Premiss bekreftet før sveip.

Multi-regel, IKKE naiv s/Azure AI Foundry/Microsoft Foundry/ — MS dropper
«Azure AI» (legger IKKE til «Microsoft») for to produktvarianter:
- «Azure AI Foundry Agent[ Service|s]» → «Foundry Agent Service/Agents» (MS-form)
- «Azure AI Foundry Models» → «Foundry Models» (i «Azure OpenAI in Foundry Models»)
- «Azure AI Foundry SDK» → «Microsoft Foundry SDK» (operatør-valg)
- «Azure AI Foundry portal/project» + generisk → «Microsoft Foundry»
- Pre-eksisterende «Microsoft Foundry Models» (4) normalisert → «Foundry Models»

Bevart: «Azure OpenAI», «Azure AI Inference SDK», «Azure AI Search»,
«Azure AI Services», kode-IDer. Historisk ref «(tidligere Azure AI Foundry)»
i model-catalog-2026.md beskyttet via lookbehind. URL /azure/ai-foundry/→
/azure/foundry/ kun i owasp-llm-top10 (KB-ref); docs/-filer deferred.

Scope: skills (inkl. 3 SKILL.md) + commands + agents + README + CLAUDE.
Ekskludert: docs/ (interne), playground/+tests/ fixtures (testdata),
CHANGELOG.md (historisk logg), STATE.md (gitignored).

3 SKILL.md endret (advisor/engineering/security) → judge-cache teknisk
invalidert for disse, men scorer uendret: advisor 91, eng/gov/infra/sec 96
(alle ≥90). validate 239/0. 0 «Azure AI Foundry» igjen (utenom bevart ref).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
 2026-06-23 21:00:27 +02:00
Kjell Tore Guttormsen
dd1036ab8a fix(ms-ai-architect): Foundry URL-navnerom-migrering (ai-foundry → foundry/foundry-classic, 141 filer) 
Task #5 del 1/3 (URL-migrering). Verifiseringen motbeviste STATE.md-premisset om ren prefix-swap: rebrand er per-URL, ikke mekanisk. En blind sed ai-foundry→foundry ville lagd 56 nye 404-er (classic-stiene finnes ikke under nytt foundry/-prefiks — bekreftet empirisk).

Metode: resolverte alle 237 unike KB-URLer mot live redirects (curl -L), bygde full-URL→full-URL-mapping fra faktisk url_effective. Bevarer locale-form, query (?view=) og #fragment per lenke.

- 231 navnerom-erstatninger over 141 filer (408 forekomster):
  - 161 → azure/foundry/ (98 ren prefix-swap + 10 sti-reorg + reorg-tilfeller)
  - 69 → azure/foundry-classic/ (eldre hub-spor: assistants, hub-DR, on-your-data; faktisk redirect-mål per operatorvalg)
  - 1 → azure/foundry-local/

- 2 døde lenker (404) fikset til verifiserte mål:
  - agent-service → azure/foundry/agents/overview
  - concepts/evaluation-evaluators/ → azure/foundry/how-to/evaluate-generative-ai-app

- 5 path-/display-referanser (uten https://, i backticks/lenketekst) rettet manuelt.

- 6 slug-baserte ai-foundry-treff urørt (scope-grense): managed-grafana-dashboard, security-baseline, power-platform prompt-builder, architecture baseline-chat (sistnevnte slug-rebrand i annet navnerom — mulig fremtidig funn).

- Parkert til task #5 del 2/3: Norway East GPT-5-datasuverenitet-fiks + modellkatalog-utvidelse (5.3/5.4/5.5, gpt-oss, sora-2).

Verifisert: 0 gjenværende azure/ai-foundry/-navnerom i skills/. validate-plugin.sh 219 PASS. test-kb-integrity.sh 117/117 passed.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Claude-Session: https://claude.ai/code/session_01REiKFhP4w6xGXXqWKpPCJJ
 2026-06-18 13:37:06 +02:00
Kjell Tore Guttormsen
84e3cfaa03 fix(ms-ai-architect): re-baseline kostnadslaget (GPT-5 $1.25/$10, én prissannhet, 10 filer) 
Audit-funn Tema B + action-items 149/150. Verifiseringen avdekket at kostnadslag-råten var bredere enn auditens 3 navngitte filer (10 filer totalt).

- GPT-5 flagship $10-15/$40-60 -> $1.25/$10.00 (verifisert juni 2026, 8-12x for hoy). gpt-5-mini $0.25/$2.00, gpt-5-nano $0.05/$0.40. Fjernet falske "Verified MCP"-stempel pa estimat-band.

- deterministic-cost-calculation-model.md gjort til ENESTE kanoniske prissannhet (§1.1). Strippet duplikat pris-tabell fra cost-models.md (rotarsak til GPT-4o $10/$30-motsetning) -> peker til registeret.

- Reconcile-funn utover audit: deterministic hadde GPT-4.1-serien HALVERT ($1/$4 og $0.20/$0.80) -> verifisert $2.00/$8.00 og $0.40/$1.60. multi-model-strategy-costs.md NOK-tabell var fabrikkert 10-47x for hoy (gpt-4o 250/1000 NOK -> 26/105). model-selection (GPT-5 $3/$12), reasoning-models (GPT-5 $20-30/$80-100), token-counting (gpt-4.1 cached $1->$0.50), small-language-models (GPT-4o output 60->105 NOK) rettet.

- Recalk alle NOK-scenarioer reproduserbart: gpt-5 ~= gpt-4.1 i kost (ikke 30-55x dyrere som gammel tekst pasto) - billigere input ($1.25 vs $2.00), dyrere output. Reell kostnadsdriver = resonnerings-tokens (fakturert som output).

- Fjernet GPT-3.5 Turbo-anbefaling (utfaset). Alignet kurs 11->10.5 NOK/USD i cost-kommando. Cost-agent eksempel: legacy GPT-4 -> GPT-4o-mini, fjernet falskt Verified-stempel.

- kb-update apply cost-bucket deferret til task #8 (kommando-feature, ikke tall-justering).

Verifisert mot OpenAI API pricing + Azure OpenAI pricing via WebSearch 2026-06-18. validate-plugin.sh 219 PASS.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Claude-Session: https://claude.ai/code/session_01REiKFhP4w6xGXXqWKpPCJJ
 2026-06-18 12:45:21 +02:00
Kjell Tore Guttormsen
baa2d0220b feat(ultraplan-local): v1.6.0 — /ultraresearch-local deep research command 
Add /ultraresearch-local for structured research combining local codebase
analysis with external knowledge via parallel agent swarms. Produces research
briefs with triangulation, confidence ratings, and source quality assessment.

New command: /ultraresearch-local with modes --quick, --local, --external, --fg.
New agents: research-orchestrator (opus), docs-researcher, community-researcher,
security-researcher, contrarian-researcher, gemini-bridge (all sonnet).
New template: research-brief-template.md.

Integration: --research flag in /ultraplan-local accepts pre-built research
briefs (up to 3), enriches the interview and exploration phases. Planning
orchestrator cross-references brief findings during synthesis.

Design principle: Context Engineering — right information to right agent at
right time. Research briefs are structured artifacts in the pipeline:
ultraresearch → brief → ultraplan --research → plan → ultraexecute.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
 2026-04-08 08:58:35 +02:00