|
|
ba597eb988
|
feat(ms-ai-architect): Sesjon 15 — B2 K10 søsken-scope-ikke-overlapp
- Refaktor: overlap-kjerne flyttet til scripts/kb-eval/lib/sibling-overlap.mjs
(bryter sirkulær import eval.mjs<->detect); detect re-eksporterer → B1-tester urørt
- K10 = søsken-scope-ikke-overlapp, deterministisk cross-skill: perSkillSiblingOverlap
+ attachSiblingOverlap i eval.mjs. combined = boundaryTension + df-vektet leksikalsk;
per-skill verdikt = verste søskenpar; terskel 7.0 (naturlig gap 7.42→6.67)
- Empirisk (alle 5): eng+infra FAIL (7.42 mot hverandre), advisor/gov/sec PASS
→ eng↔infra-signal mater B3 merge/saner (ikke blokkering)
- Gated baseline-regen via --write (descriptions urørt → judge K1/K4/K7/K8/K9 merget
uendret, ikke fabrikkert); rubric K1-K10
- TDD: +9 tester (tests/kb-eval/test-k10-sibling-overlap.test.mjs), kb-eval 31→40
- 0 skriving til skills/. Suiter: validate 239 · kb-update 122 · kb-integrity 192/192
|
2026-06-20 11:59:59 +02:00 |
|
|
|
c778cf6cd2
|
fix(ms-ai-architect): Sesjon 12 — lukk SPOR A, ny eval-baseline (K1–K9 grønn alle 5)
Regenerert eval-baseline.json via gated `eval.mjs --write` så baselinen fanger
S9+S10+S11 judge-verdiktene (var pre-S9). Eneste endrede artefakt: baseline-fila
(123+/128−, ren regenerering — judge-felt null→populert fra data/judge-results.json).
Verifisert FØR skriving (read-only) og i resultatet:
- judge-cache: K1/K4/K7/K8/K9 PASS alle 5 (advisor K8=0.8 at-threshold, pass)
- deterministisk: K2/K3/K5/K6/refTall PASS alle 5
- ny baseline: 0 åpen FAIL K1–K9 på alle 5 skills, judge populert alle 5
SPOR A (krav 1–4) er nå komplett: K1 autoritativ+blindet (S11), 6×5 kanonisk (S8),
K4+K9 body-cleanup (S9 sec+gov, S10 eng+infra), skill-optimalisering refTall+K5 (S7).
eval-baseline.json er sannhetskilden og grønn på K1–K9.
Ingen regresjon: validate 239 · kb-update 122 · kb-eval 15 · kb-integrity 192/192.
NESTE: Spor B (scheduled deteksjon · kurs-spor #25 · low-currency · onboarding #26).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01REiKFhP4w6xGXXqWKpPCJJ
|
2026-06-20 10:11:45 +02:00 |
|
|
|
215772df87
|
feat(ms-ai-architect): Sesjon 1 - skill eval-baseline (rubrikk K1-K9) + ai-foundry probe
Read-only eval-verktoy scripts/kb-eval/eval.mjs (determ. K2/K3/K5/K6 + ref-tall) + operator-gated LLM-judge (judge-prompt.md -> data/judge-results.json) flettet til data/eval-baseline.json. 13 enhetstester (tests/kb-eval/), 42 kb-update-tester gronne. Baseline: K5 3/5 fail, K9 4/5 fail, K4 2/5 fail, ref-tall 2/5 fail; 2 konkrete bugs flagget (security 6x5-vekting-motstrid, governance broken ref-path).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01REiKFhP4w6xGXXqWKpPCJJ
|
2026-06-19 20:26:56 +02:00 |
|
