Conversation
Eval accuracy per key field from the committed baseline (replay, 15 synthetic cases), the recorded state of both sample requests, three key decisions and the limits; handling time marked as not yet measured, as decided by the orchestrator on 2026-09-28. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01DJ5vaKvTYiMvdngT4d3xo1
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01DJ5vaKvTYiMvdngT4d3xo1
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
…s, ADR links (#107 review) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01DJ5vaKvTYiMvdngT4d3xo1
Owner
Author
Frischer Review (read-only Subagent, Faktencheck) – Ergebnis und UmgangNichts blockierend. Bestätigt: alle Tabellenwerte =
Generated by Claude Code |
Fluory
marked this pull request as ready for review
October 2, 2026 08:28
This was referenced Oct 2, 2026
This branch had an error being deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Warum
Fixes #76 · Showcase-Review 2026-09-27, P1 Punkt 4 (Epic #19) · Umfang laut Entscheidung des Orchestrators vom 2026-09-28: veröffentlichen, was messbar ist; Bearbeitungszeit ausdrücklich „noch nicht gemessen“.
Arbeitsstand
docs/product/evidence.md(Eval-Genauigkeit je Schlüsselfeld, reproduziert mitpnpm evals; Zustand beider Musterfälle aus den Aufzeichnungen gezählt; drei Kernentscheidungen mit Links auf ADR-0001; Grenzen), Link im README, CHANGELOG; frischer Review (Faktencheck) eingearbeitet; CI grün.Was ist passiert (Klartext)
Es gibt jetzt eine Seite mit nachprüfbaren Zahlen: wie genau die Verarbeitung je Feld auf 15 synthetischen Testfällen arbeitet, was bei den zwei Musteranfragen erkannt wurde und was eine Person noch entscheiden muss, und welche drei Entscheidungen das Produkt tragen. Jede Zahl ist als synthetisch gekennzeichnet und lässt sich mit einem Befehl nachrechnen. Wichtig und ehrlich benannt: Die Testfälle nutzen vorbereitete KI-Antworten – sie messen, wie gut unsere Prüfung arbeitet, nicht wie gut ein Live-Modell ist. Die Zeitersparnis ist noch nicht gemessen und wird deshalb auch nicht behauptet.
Plan-Pflicht (SYSTEM.md §4)
Geändert
docs/product/evidence.mdREADME.md: Link in der Doku-TabelleCHANGELOG.md: Added-Eintrag docs(showcase): publish measured evidence – field accuracy, corrections, handling time #76Nachweis (SYSTEM.md §11)
pnpm evalsam 2026-10-02 auf606caec→ identisch mitservices/ai/evals/baseline.json(Gate PASSED, 15 Fälle)src/features/samples/data/*.recording.json(18eab97,d51ad56): Werk Ost Kopf 4/1/1 + 15 Positionswerte gefunden; Pumpe P204 Kopf 6/0/0 + 10 Positionswerte gefunden; Grund aus der Aufzeichnung (calendar_week_only)verify:changed: grün –scripts/doku-check.shverify: grün – CIcheckauf 4aeea63 (Doku-only; der Code-Teil vonverifyist unberührt)verify:full/ E2E-Spec: nicht betroffenDoku-Entscheidung (genau eine)
docs/product/evidence.md, README-Link[Unreleased](sichtbares Feature oder Verhalten – im selben PR, nie „später"): Added-Eintrag docs(showcase): publish measured evidence – field accuracy, corrections, handling time #76Entferntes oder Umbenanntes: nichts entfernt
Dateigrößen und neue Bausteine (SYSTEM.md §7)
Dateien über 500 Zeilen im Diff (Ausnahmen: generierter Code, Lockfiles, Fixtures, Migrationen, Schemas, Ressourcen, Doku, Konfiguration):
Über 800 Zeilen mit neuer Fachlogik oder über 1000 Zeilen (P1/P2): nicht betroffen
Neue Shared-Komponente, Utility-Datei, Adapter oder fachlicher Service:
Subagent-Einsätze
Risiken / offene Punkte
🤖 Generated with Claude Code
https://claude.ai/code/session_01DJ5vaKvTYiMvdngT4d3xo1