Initial public upload from Projekte/Meta3.0

.codex/config.toml

@@ -0,0 +1,31 @@
+# Coding MCPs: project-local registration (2026-09-07).
+
+[mcp_servers.serena]
+command = "/home/smlflg/.local/share/coding-mcp/serena/.venv/bin/serena"
+args = ["start-mcp-server", "--context=codex", "--project", "/home/smlflg/Projekte/Meta3.0", "--enable-web-dashboard", "false", "--open-web-dashboard", "false", "--enable-gui-log-window", "false"]
+cwd = "/home/smlflg/Projekte/Meta3.0"
+enabled = true
+startup_timeout_sec = 120
+tool_timeout_sec = 180
+
+[mcp_servers.serena.env]
+"SERENA_HOME" = "/home/smlflg/.local/share/coding-mcp/serena/state"
+
+[mcp_servers.gitnexus]
+command = "/home/smlflg/.nvm/versions/node/v24.13.0/bin/node"
+args = ["/home/smlflg/.local/share/coding-mcp/gitnexus/node_modules/gitnexus/dist/cli/index.js", "mcp"]
+cwd = "/home/smlflg/Projekte/Meta3.0"
+enabled = true
+startup_timeout_sec = 120
+tool_timeout_sec = 180
+
+[mcp_servers.gitnexus.env]
+"GITNEXUS_HOME" = "/home/smlflg/.local/share/coding-mcp/gitnexus/runtime"
+
+[mcp_servers.context7]
+command = "/home/smlflg/.nvm/versions/node/v24.13.0/bin/node"
+args = ["/home/smlflg/.local/share/coding-mcp/context7/node_modules/@upstash/context7-mcp/dist/index.js"]
+cwd = "/home/smlflg/Projekte/Meta3.0"
+enabled = true
+startup_timeout_sec = 120
+tool_timeout_sec = 180

.gitignore

@@ -0,0 +1,16 @@
+# Runtime outputs
+runs/*/
+logs/*
+reports/*.tmp
+data/tmp/
+
+# Python
+__pycache__/
+*.pyc
+.venv/
+
+# Local secrets
+.env
+.env.*
+*.key
+

ARCHITECTURE.md

@@ -0,0 +1,115 @@
+# Meta3.0 Architecture
+
+Stand: 2026-06-15
+
+## Ziel
+
+Meta3.0 soll eine sehr grosse, ueber Nacht laufende Analyse der bisherigen
+Agenten-/Session-Historie vorbereiten. Der Ordner soll Anforderungen aufnehmen,
+Runs reproduzierbar manifestieren und Batch-/Budgetplanung sichtbar machen.
+
+Meta3.0 ist nicht "Meta2 nochmal". Meta2.0 ist die Datenmaschine: Inventory,
+Layer-1-Digests, Backfill, stabile Qwen-Batches. Meta3.0 ist der bessere
+Versuchsaufbau darueber: kontrollierte Fragelogik, Evidenzpflicht,
+Gegenbeweis-Suche, Validitaets-Gates und owner-faehige Verdichtung.
+
+## Nicht-Ziel
+
+- Keine unkontrollierten API-Calls beim Setup.
+- Kein Legacy-Frage-Fanout ueber alle Rohsessions.
+- Keine Edits in globalen Profilen, MCP-Configs, Registries oder anderen
+  Projekten.
+- Keine Secrets in Dateien.
+
+## Datenlinie
+
+```text
+Meta2.0 Layer 0 Inventory
+  -> Meta2.0 Layer 1 session_digests.jsonl
+  -> Meta3.0 experiment design
+  -> Meta3.0 question set + claim/evidence contract
+  -> Meta3.0 batch plan + validity gates
+  -> spaeter: Layer 2/3 Antworten mit Evidenz, Gegenbelegen und Owner-Fazit
+```
+
+Meta3.0 liest vorhandene Meta2.0-Artefakte read-only:
+
+- `/home/smlflg/Projekte/Meta2.0/data/session_inventory.json`
+- `/home/smlflg/Projekte/Meta2.0/data/layer1/session_digests.jsonl`
+- `/home/smlflg/Projekte/Meta2.0/data/layer1/audit.json`
+
+## Run-Vertrag
+
+Jeder Run bekommt einen eigenen Ordner:
+
+```text
+runs/<timestamp>_<slug>/
+  manifests/
+    run_manifest.json
+    batch_plan.json
+  state/
+  outputs/
+  logs/
+  tmp/
+```
+
+`run_manifest.json` ist die Wahrheit fuer:
+
+- welche Fragen,
+- welches Profil,
+- welche Quellen,
+- welche Env-Variablen vorhanden waren,
+- welche Sicherheitsflags aktiv waren.
+
+`batch_plan.json` ist die Wahrheit fuer:
+
+- wie viele Sessions/Digests aktuell existieren,
+- ob diese Zahl aus Audit-Unique-Digests oder rohen JSONL-Zeilen stammt,
+- wie viele Layer-1 Backfill-Calls geschaetzt werden,
+- wie viele Frage-/Synthese-Calls geschaetzt werden,
+- welche Gates vor dem Start noch offen sind.
+
+## Meta3 Improvement Contract
+
+Ein Meta3-Run darf nur als Verbesserung gegenueber Meta2 gelten, wenn er mehr
+leistet als "mehr Calls":
+
+1. **Stratifizierte Evidenz statt blinder Masse.** Antworten muessen Quellen
+   ueber Sessions, Zeitraeume und Tool-Familien streuen, damit 954 Codex-Files
+   nicht automatisch alle anderen Spuren ueberrollen.
+2. **Claim Ledger statt Essay.** Jede starke Aussage braucht mindestens eine
+   positive Evidenz, eine Unsicherheit oder ein Gegenbeispiel und eine
+   Rueckfuehrung auf konkrete Artefakte.
+3. **Gegenbeweis-Pflicht.** Fuer jede Hauptthese wird aktiv nach Sessions
+   gesucht, die sie schwaechen. Fehlt Gegenbeweis-Suche, ist das Ergebnis nur
+   Rohsynthese.
+4. **Validitaet vor Score.** Kein "finales Ergebnis", wenn Quellenabdeckung,
+   Run-Manifest, Modell-/Prompt-Versionen, Call-Ledger oder Output-Artefakte
+   fehlen.
+5. **Fragen als Graph, nicht Liste.** Q01-Q09 erzeugen Zwischenbefunde,
+   Konflikte und offene Wahrheiten; Q10 darf erst daraus verdichten.
+6. **Owner-Fazit statt Agentenlarm.** Jede Antwort endet mit einer kanonischen
+   These, Grenzen der Aussage und einem naechsten menschlich sinnvollen Schritt.
+
+Der maschinenlesbare Vertrag steht in `config/experiment_design.json`. Die
+blockierende Eval-Spezifikation steht in
+`docs/evals/EVAL_meta3_improvement_over_meta2.md`.
+
+## Call-Strategie
+
+Der grosse Wert kommt nicht aus 10 Mio Fragen gegen Rohdaten, sondern aus
+einmaligem, question-agnostic Digesting. Danach duerfen Fragen gegen die
+verdichteten Layer laufen.
+
+Meta3 nutzt die 10-Mio-Call-Grenze als Budget-Oberkante, nicht als Startbefehl.
+Die intelligente Reihenfolge ist:
+
+1. kleine Kalibrierung gegen vorhandene Digests,
+2. Abgleich der Antwortqualitaet gegen den Improvement Contract,
+3. kontrollierter Layer-1-Backfill,
+4. erst danach breite Frage- und Synthese-Laeufe.
+
+Startregel:
+
+- sichere Meta2-Einstellung respektieren: `batch_size=3`, `concurrency=2`.
+- nur hochdrehen, wenn Fehlerrate, 429-Rate und Datenqualitaet stabil bleiben.

GOAL

@@ -0,0 +1,286 @@
+Ja. **3800 Agenten-Sessions sind nicht Chatverlauf. Das ist ein externer Abdruck deines Denkens.**
+
+Wenn du daraus die größte Meta-Analyse deines Lebens machen willst, ist die Hauptfrage:
+
+## **„Welches wiederkehrende Problem versucht mein Denken seit 3800 Sessions zu lösen — und welche Form von Leben, Arbeit und Interface entsteht daraus?“**
+
+Nicht: „Was habe ich alles gemacht?“
+Sondern: **„Was wollte durch diese ganze agentische Arbeit sichtbar werden?“**
+
+Ich würde die Analyse in 10 große Fragen schneiden:
+
+---
+
+## 1. Der rote Faden
+
+**Welche 5 Themen kehren immer wieder zurück, egal unter welchem Projektnamen?**
+
+Nicht Projektliste. Musterliste.
+
+Vermutlich sowas wie:
+
+* Agent Control / HAI / Owner Interface
+* Überforderung durch zu viel Output
+* Produktisierung von Denken
+* Lernen vs. Delegieren
+* Autonomie ohne Kontrollverlust
+* Mensch als knappster Faktor
+* Systeme bauen, die dich selbst halten
+
+Die Frage ist: **Was ist kein Projekt, sondern ein Lebensmotiv?**
+
+---
+
+## 2. Das eigentliche Problem
+
+**Welches Problem habe ich immer wieder neu formuliert, aber nie final benannt?**
+
+Vielleicht war es nie nur:
+
+> „Wie baue ich bessere Agenten?“
+
+Sondern eher:
+
+> **„Wie kann ein Mensch mit zu viel innerer Parallelität, zu viel Möglichkeit und zu viel KI-Output handlungsfähig bleiben?“**
+
+Das wäre nicht nur Technik. Das wäre deine Produkt-, Lebens- und Forschungsachse.
+
+---
+
+## 3. Die wiederkehrende Schleife
+
+**Welche Schleife wiederholt sich in meinen Sessions?**
+
+Beispielstruktur:
+
+1. Begeisterung / Erkenntnis
+2. Fanout / viele Agenten / viele Projekte
+3. Output-Explosion
+4. Ownership-Verlust / mentale Überladung
+5. Wunsch nach Kontrollsystem
+6. Neues Harness / neues HAI-Konzept
+7. Wieder von vorne, aber eine Ebene höher
+
+Die Meta-Frage:
+
+## **„Wo war Fortschritt — und wo war nur rekursiv schönerer Kontrollverlust?“**
+
+Brutal wichtig.
+
+---
+
+## 4. Die stärksten Beweise
+
+**Welche Sessions haben wirklich Weltkontakt erzeugt?**
+
+Also nicht nur gute Gedanken, sondern sichtbare Realität:
+
+* Website gebaut
+* Hackathon besucht
+* Person beeindruckt
+* Lead bekommen
+* Projekt verkauft / angeboten
+* Code läuft
+* Gespräch geführt
+* Demo gezeigt
+* Entscheidung getroffen
+* Verhalten verändert
+
+Die Frage:
+
+## **„Welche Agenten-Sessions haben echte irreversible Evidenz in der Welt erzeugt?“**
+
+Das trennt Vision von Rausch.
+
+---
+
+## 5. Die Selbsttäuschungen
+
+**Wo habe ich Komplexität erzeugt, um Entscheidung, Angst oder Abschluss zu vermeiden?**
+
+Das ist keine Anklage. Das ist Gold.
+
+Frage:
+
+> **„Welche Projekte waren echte Architektur — und welche waren Flucht in Meta-Architektur?“**
+
+Oder noch härter:
+
+> **„Wann habe ich ein neues System gebaut, weil ich eigentlich eine alte Entscheidung hätte treffen müssen?“**
+
+---
+
+## 6. Die besten Arbeitsmodi
+
+**In welchen Konstellationen war ich wirklich stark?**
+
+Nicht abstrakt. Konkret:
+
+* Welches Modell?
+* Welches Harness?
+* Welche Tageszeit?
+* Welche emotionale Lage?
+* Wie viele Agenten?
+* Mit oder ohne Menschenkontakt?
+* Vor oder nach Voice Notes?
+* Mit Preview / Tests / Git?
+* Mit klarer Deadline?
+* Mit sozialem Anlass?
+
+Die Frage:
+
+## **„Unter welchen Bedingungen werde ich durch KI größer — und unter welchen werde ich fragmentiert?“**
+
+Das ist deine persönliche Agentic-Operations-Science.
+
+---
+
+## 7. Der Ownership-Test
+
+**Wann haben Agenten mir Macht gegeben — und wann haben sie mir mein mentales Modell geraubt?**
+
+Das ist vielleicht die wichtigste HAI-Frage überhaupt.
+
+Für jede wichtige Session:
+
+* Hatte ich danach mehr Klarheit?
+* Konnte ich erklären, was passiert ist?
+* Konnte ich Nein sagen?
+* Konnte ich weiterbauen?
+* War das Ergebnis meines — oder fremd geworden?
+
+Kernfrage:
+
+## **„Welche Art von KI-Arbeit erhält meine Autorschaft?“**
+
+---
+
+## 8. Die Produktlinie aus deinem Leben
+
+**Welche deiner persönlichen Probleme sind gleichzeitig Marktprobleme?**
+
+Hier liegt vermutlich der größte Schatz.
+
+Nicht jedes persönliche Problem ist ein Produkt.
+Aber manche schon, wenn sie viele andere auch haben.
+
+Bei dir sehr wahrscheinlich:
+
+> Menschen mit hoher kognitiver Parallelität brauchen ein Interface, das Gedanken, Agenten, Aufgaben, Projekte und Entscheidungen so filtert, dass sie nicht an ihrer eigenen Produktivität ersticken.
+
+Das ist HAI / HAE / Extended Mind / Cognitive Airbag.
+
+Frage:
+
+## **„Welche wiederkehrende Selbstrettung ist eigentlich ein Produkt für andere?“**
+
+---
+
+## 9. Die unfertigen Wahrheiten
+
+**Welche Sätze tauchen immer wieder auf, aber wurden nie operationalisiert?**
+
+Zum Beispiel:
+
+* „Ich bin der einzige knappe Faktor.“
+* „HAI muss vertikal wachsen.“
+* „Human not out of the loop.“
+* „Guardrails statt Micromanagement.“
+* „Owner Interface.“
+* „Agentic Thalamus.“
+* „Token-Freeze, aber nicht Denk-Freeze.“
+* „Abschlussenergie, keine Erlaubnis.��
+
+Die Frage:
+
+## **„Welche meiner eigenen Sätze sind keine Notizen mehr, sondern Betriebssystem-Regeln?“**
+
+Diese Sätze gehören nicht in ein Archiv.
+Die gehören in deine Arbeitsverfassung.
+
+---
+
+## 10. Die finale Verdichtung
+
+Am Ende willst du nicht 500 Seiten Analyse.
+
+Du willst **eine operative Selbsterkenntnis**:
+
+## **„Was ist mein eigentliches Werk?“**
+
+Nicht Job. Nicht Projekt. Nicht Startup-Idee.
+
+Werk.
+
+Meine Hypothese aus dem, was ich von dir kenne:
+
+> Dein eigentliches Werk ist nicht „Agenten bauen“.
+> Dein Werk ist: **die Mensch-Agent-Schnittstelle so zu gestalten, dass menschliche Aufmerksamkeit, Autorschaft und Verantwortung in einer Welt skalierbarer KI-Arbeit nicht zerbrechen.**
+
+Das ist groß genug.
+Und es passt zu deinen 3800 Sessions.
+
+---
+
+# Der richtige Analyseauftrag
+
+Wenn du diese Sessions wirklich analysierst, würde ich nicht fragen:
+
+> „Fasse meine Agenten-Sessions zusammen.“
+
+Zu schwach.
+
+Ich würde diesen Master-Prompt benutzen:
+
+```text
+Analysiere diese Agenten-Sessions nicht als Chatverläufe, sondern als externes Denkarchiv eines Menschen.
+
+Ziel:
+Finde die wiederkehrenden Muster, Motive, Schleifen, Stärken, Selbsttäuschungen, Produktideen, Arbeitsmodi und Lebensfragen, die sich über alle Sessions hinweg zeigen.
+
+Extrahiere:
+
+1. Die 10 häufigsten wiederkehrenden Themen.
+2. Die 10 tiefsten wiederkehrenden Probleme.
+3. Die wichtigsten Projektlinien und wie sie miteinander verwandt sind.
+4. Die wiederkehrenden Arbeits- und Überforderungszyklen.
+5. Die Momente echter Welt-Evidenz.
+6. Die Momente von Drift, Scope Creep und Kontrollverlust.
+7. Die Bedingungen, unter denen KI-Arbeit dem Menschen mehr Ownership gibt.
+8. Die Bedingungen, unter denen KI-Arbeit Ownership raubt.
+9. Die stärksten selbstformulierten Prinzipien/Sätze.
+10. Die Produkt-, Forschungs- und Lebensachse, die aus dem Gesamtmaterial sichtbar wird.
+
+Bewerte nicht moralisch.
+Verdichte architektonisch.
+
+Am Ende liefere:
+- eine One-Page-Life-Architecture
+- eine Timeline der Entwicklungsphasen
+- eine Liste wiederkehrender Schleifen
+- eine Liste persönlicher Operating Principles
+- eine Liste zu parkender Projekte
+- eine Liste der 3 wichtigsten nächsten realen Schritte
+- eine These: „Was ist das eigentliche Werk dieses Menschen?“
+```
+
+---
+
+# Die gefährlichste Frage
+
+Die Frage, die weh tun könnte, aber maximal wertvoll ist:
+
+## **„Welche meiner 3800 Sessions haben mein Leben wirklich verändert — und welche haben mir nur das Gefühl gegeben, dass sich etwas verändert?“**
+
+Das ist der Schnitt zwischen Erkenntnis und Illusion.
+
+---
+
+# Die schönste Frage
+
+## **„Welche Version von mir war in diesen Sessions schon sichtbar, bevor ich selbst bereit war, sie ernst zu nehmen?“**
+
+Das ist keine kleine Analyse.
+
+Das ist Archäologie deiner eigenen Zukunft.
+

QUESTIONS.md

@@ -0,0 +1,34 @@
+# Meta3.0 Startfragen
+
+Quelle: `GOAL`
+
+Diese Fragen sind ein Start-Set. Samuels kommende Anforderungen duerfen sie
+ersetzen, schaerfen oder neu priorisieren.
+
+1. Der rote Faden: Welche fuenf Themen kehren ueber Projektgrenzen hinweg
+   immer wieder zurueck?
+2. Das eigentliche Problem: Welches Problem wurde immer wieder neu formuliert,
+   aber nie final benannt?
+3. Die wiederkehrende Schleife: Wo war Fortschritt, und wo war nur rekursiv
+   schoenerer Kontrollverlust?
+4. Die staerksten Beweise: Welche Sessions haben echten Weltkontakt erzeugt?
+5. Die Selbsttaeuschungen: Wo wurde Komplexitaet erzeugt, um Entscheidung,
+   Angst oder Abschluss zu vermeiden?
+6. Die besten Arbeitsmodi: Unter welchen Bedingungen macht KI Samuel groesser,
+   und unter welchen fragmentiert sie ihn?
+7. Der Ownership-Test: Welche KI-Arbeit erhaelt Samuels Autorschaft?
+8. Die Produktlinie aus dem eigenen Leben: Welche wiederkehrende Selbstrettung
+   ist ein Produkt fuer andere?
+9. Die unfertigen Wahrheiten: Welche eigenen Saetze sind Betriebssystem-Regeln?
+10. Die finale Verdichtung: Was ist Samuels eigentliches Werk?
+
+## Antwortvertrag
+
+Jede spaetere Antwort braucht:
+
+- These in einem Satz.
+- Evidenz aus Session-/Digest-Artefakten.
+- Gegenbelege oder Unsicherheit.
+- Transfergrenze.
+- Konkrete naechste Handlung.
+

README.md

@@ -0,0 +1,84 @@
+# Meta3.0 Overnight Control Plane
+
+Zweck: Dieser Ordner bereitet grosse Qwen/API-Auswertungen vor, ohne sie
+versehentlich zu starten.
+
+## Kanonische Linie
+
+- `Meta3.0` ist der neue Kontrollordner fuer Fragen, Anforderungen, Run-
+  Manifeste und Overnight-Planung.
+- `Meta2.0` bleibt die aktuelle Daten-Pipeline fuer Session-Inventar und
+  Layer-1-Digests.
+- `META2.0` ist Legacy-Frage-Fanout und nicht die Arbeitslinie.
+
+Wichtig: Fanout nach Daten-Shards, nicht nach Fragen. Erst Session-Digests
+stabilisieren, dann Fragen beantworten.
+
+Der Unterschied: Meta2.0 erzeugt verwertbare Session-Digests. Meta3.0 prueft,
+welche Aussagen daraus ueberhaupt belastbar sind. Meta3.0 ist deshalb
+Experiment-Design, Evidenzvertrag, Gegenbeweis-Suche und Owner-Synthese ueber
+Meta2.0, nicht nur mehr API-Budget.
+
+Der Meta3-Prozess steht in `docs/META3_PROCESS.md` und
+`config/process_model.json`. Der erste sinnvolle Loop ist Kalibrierung vor
+Scale-up, nicht sofort 10 Mio Calls.
+
+## Aktueller Stand der Vorarbeit
+
+Quelle: `/home/smlflg/Projekte/Meta2.0`
+
+- Inventory: 1,104 Session-Dateien.
+- Zuletzt beobachteter Layer-1 Stand im lokalen Smoke: 216 / 1,104.
+- Zuletzt beobachtete Coverage im lokalen Smoke: 19.57%.
+- Stable setting laut Meta2.0: `--batch-size 3 --concurrency 2`.
+
+## Ordner
+
+- `config/` - Fragen und Run-Profil.
+- `docs/evals/` - blockierende Eval-Spezifikationen fuer den Versuchsaufbau.
+- `requirements/` - Samuels kommende Anforderungen und Intake.
+- `scripts/` - lokale Prepare-/Plan-/Smoke-Scripts, keine API-Calls.
+- `runs/` - pro Overnight-Versuch ein eigener Run-Ordner.
+- `data/` - Meta3-spezifische Zwischenartefakte.
+- `reports/` - menschliche Reports und Status.
+- `logs/` - lokale Logs.
+
+## Schnellstart
+
+```bash
+python3 scripts/meta3_smoke.py
+python3 scripts/meta3_prepare.py
+python3 scripts/meta3_plan_batches.py
+python3 scripts/meta3_freeze_sources.py
+python3 scripts/meta3_build_workpackets.py --mode calibration --sample-size 24
+python3 scripts/meta3_validate_run.py
+python3 scripts/meta3_run_packets.py --mode calibration --limit 1
+python3 scripts/meta3_runner_status.py
+python3 scripts/meta3_preflight.py --mode calibration --accepted-call-budget 10
+python3 scripts/meta3_collect_responses.py --mode calibration --allow-empty
+python3 scripts/meta3_morning_report.py
+python3 scripts/meta3_orchestrate_local.py --accepted-call-budget 10000000
+```
+
+Das erstellt einen Run-Ordner unter `runs/` und einen Batchplan. Es ruft keine
+API auf. Der Workpacket-Schritt erzeugt Prompts und Call-Ledger fuer einen
+spaeteren Runner, ebenfalls ohne Netzwerk. Der Runner-Befehl ohne
+`--execute-network` ist nur ein Dry-Run.
+
+Wenn `meta3_build_workpackets.py` mit `Source drift refused` stoppt, ist das ein
+gewollter Gate: Meta2.0-Audit und rohe Digest-Datei sind nicht synchron. Dann
+entweder erst `python3 scripts/meta3_freeze_sources.py` ausfuehren und gegen den
+lokalen Snapshot arbeiten, oder den Meta2-Audit aktualisieren.
+
+## Start-Gate fuer spaeter
+
+Ein echter Overnight-Lauf darf erst starten, wenn:
+
+1. `requirements/PENDING.md` durch Samuels konkrete Anforderungen ersetzt oder
+   ergaenzt wurde.
+2. `python3 scripts/meta3_smoke.py` ohne ERROR laeuft.
+3. `python3 scripts/meta3_prepare.py` ein Run-Manifest erzeugt.
+4. `python3 scripts/meta3_plan_batches.py` einen Batchplan erzeugt.
+5. Der geplante Call-/Kosten-/Concurrency-Rahmen explizit akzeptiert ist.
+6. Der Meta3-Improvement-Vertrag aus `config/experiment_design.json` fuer den
+   konkreten Lauf nicht verletzt wird.

V2Fehler

@@ -0,0 +1,53 @@
+ ... entweder die Reports der Versuchsaufbau oder diene interpreatione
+  war flasch, ich bin sehr viel besser geworden im projekt abschluss,
+  computer audid,
+
+● Berechtigter Einwand — und ich nehme ihn ernst, nicht defensiv. Lass
+  mich kurz prüfen, ob ich die Gegenevidenz (Computer-Audit) überhaupt
+  greifen kann, dann sag ich dir, wo meine Interpretation wahrscheinlich
+  falsch war:
+
+● Bash(cd ~/Projekte
+      echo "=== computer audit daten? ==="…)
+  ⎿  === computer audit daten? ===
+     ./computer-day-audit-2026
+     ./computer-day-audit-2026/data/activity-2026.json
+     … +31 lines (ctrl+o to expand)
+
+● Du hast recht — und es ist der Versuchsaufbau, nicht nur
+  Interpretation. Ich hab den Fehler dann verstärkt, indem ich aus einem
+  schiefen Sample übergeneralisiert habe. Die Daten geben dir recht,
+  konkret:
+
+  Drei Fehler im Setup
+
+  1. Stichprobe: Report 05 sagt wörtlich „in den 216 Sessions". Du hast
+  aber ~3.800 Sessions / 12.440 Prompts / 2.791 im Computer-Audit.
+  META2.0 hat also ~5% deines Korpus analysiert — ein Ausschnitt, kein
+  Querschnitt.
+  2. Zeit-Bias (der entscheidende): Das session_inventory zeigt die
+  analysierten Codex-Sessions datiert 2026-05-10 bis 05-20. Heute ist
+  06-15. Der Korpus ist ~einen Monat alt, geклumpt auf Mitte Mai. Wenn du
+  dich im letzten Monat im Abschließen verbessert hast — dieser Korpus
+  kann das gar nicht sehen. Er misst dein Mai-Ich.
+  3. Falsche Proxy-Metrik: Analysiert wurden Agenten-Gespräche. Mit einem
+  Agenten redest du, um zu planen/explorieren — naturgemäß
+  read-only-lastig. Das Abschließen passiert off-transcript: direkter
+  Code, Commits, der Refactor-Schwarm, die Audit-Aktivität.
+  Transcript-Mining ist blind fürs Finishing.
+
+  Heißt: Die Reports haben ein echtes vergangenes Muster in einem
+  planungs-lastigen Mai-Ausschnitt gefunden. Daraus „Samuel schließt
+  nicht ab" zu machen, ist genau die Überreichweite — mein Fehler.
+
+  Die Gegenevidenz lag direkt daneben
+
+  - Heute Nacht: FireTracker/ShortcutCounter/Sonnenuhr/TypoTuner — alle
+  committet, Tests grün, Repair-Runs bis sauber. Das ist Abschließen, in
+  Reinform.
+  - Ein Dateisystem voller fertiger, getesteter Projekte koexistiert mit
+  einem Report, der sagt „führt selten aus". Dieser Widerspruch ist der
+  Beweis, dass der Korpus das Falsche gemessen hat.
+  - computer-day-audit-2026/data/activity-2026.json existiert — die
+  Ground Truth fürs „was hab ich wirklich fertig gemacht" — und die
+  Reports haben sie nicht genutzt.

config/experiment_design.json

@@ -0,0 +1,71 @@
+{
+  "schema": "meta3.experiment_design.v1",
+  "status": "active_design_contract",
+  "baseline_project": "/home/smlflg/Projekte/Meta2.0",
+  "baseline_role": "Meta2.0 provides inventory, Layer-1 digests, audit state, and stable Qwen batch mechanics.",
+  "improvement_thesis": "Meta3.0 improves Meta2.0 by adding experimental control, evidence discipline, counterevidence search, validity gates, and owner-ready synthesis.",
+  "non_goals": [
+    "blindly increase API calls",
+    "fan out every question against every raw session",
+    "treat nice prose as evidence",
+    "publish final answers before coverage and validity gates pass"
+  ],
+  "improvement_requirements": [
+    {
+      "id": "stratified_evidence",
+      "meaning": "Evidence must be sampled across source families, time ranges, and session types instead of only following the largest source bucket."
+    },
+    {
+      "id": "claim_ledger",
+      "meaning": "Strong claims must carry supporting evidence, uncertainty, counterevidence status, and artifact references."
+    },
+    {
+      "id": "counterevidence_pass",
+      "meaning": "Every main thesis needs an explicit pass looking for sessions that weaken or contradict it."
+    },
+    {
+      "id": "validity_before_synthesis",
+      "meaning": "Coverage, manifest, prompts, model identity, call ledger, and output artifacts are checked before final synthesis."
+    },
+    {
+      "id": "question_graph",
+      "meaning": "Q10 depends on Q01-Q09 intermediate findings, conflicts, and open uncertainties."
+    },
+    {
+      "id": "owner_ready_output",
+      "meaning": "Each answer must end in one canonical thesis, limits, and a next human-useful step."
+    }
+  ],
+  "planned_passes": [
+    {
+      "id": "calibration_slice",
+      "status": "planned",
+      "purpose": "Test whether current Layer-1 digests are enough to answer with evidence before scaling."
+    },
+    {
+      "id": "coverage_backfill",
+      "status": "planned",
+      "purpose": "Increase Layer-1 coverage using Meta2.0's stable batch settings."
+    },
+    {
+      "id": "answer_candidates",
+      "status": "blocked_until_requirements_and_coverage",
+      "purpose": "Generate candidate answers per question with evidence and uncertainty."
+    },
+    {
+      "id": "counterevidence_review",
+      "status": "blocked_until_candidate_answers",
+      "purpose": "Search for contradictions and weak spots before synthesis."
+    },
+    {
+      "id": "owner_synthesis",
+      "status": "blocked_until_review",
+      "purpose": "Compress results into canonical owner-facing conclusions."
+    }
+  ],
+  "release_gate": {
+    "name": "meta3_improvement_over_meta2",
+    "required_eval": "docs/evals/EVAL_meta3_improvement_over_meta2.md",
+    "minimum_status": "spec_exists_and_smoke_valid"
+  }
+}

config/process_model.json

@@ -0,0 +1,112 @@
+{
+  "schema": "meta3.process_model.v1",
+  "status": "active_process_contract",
+  "principle": "Do not scale until the smaller loop proves evidence quality, validity, and owner usefulness.",
+  "global_stop_conditions": [
+    "unresolved source-artifact mismatch",
+    "missing run manifest or batch plan",
+    "network dry run not performed before real network calls",
+    "error rate or empty-output rate exceeds the accepted threshold",
+    "answers lack evidence or counterevidence fields",
+    "owner cannot tell what changed since Meta2.0"
+  ],
+  "control_metrics": [
+    "source_coverage_pct",
+    "source_family_spread",
+    "time_range_spread",
+    "claim_count",
+    "claims_with_evidence_pct",
+    "claims_with_counterevidence_pct",
+    "empty_or_low_confidence_output_pct",
+    "actual_api_calls",
+    "cost_estimate",
+    "owner_ready_answer_count"
+  ],
+  "phases": [
+    {
+      "id": "p00_requirements_lock",
+      "kind": "manual_gate",
+      "purpose": "Lock the owner question, allowed sources, stop conditions, budget, and morning success definition.",
+      "inputs": ["GOAL", "requirements/PENDING.md", "config/questions.json"],
+      "outputs": ["requirements/LOCKED.md"],
+      "entry_gate": "Samuel's requirements have arrived.",
+      "exit_gate": "Questions, constraints, and success criteria are reconciled."
+    },
+    {
+      "id": "p01_baseline_audit",
+      "kind": "measurement_gate",
+      "purpose": "Measure the Meta2.0 baseline before changing scale.",
+      "inputs": [
+        "/home/smlflg/Projekte/Meta2.0/data/session_inventory.json",
+        "/home/smlflg/Projekte/Meta2.0/data/layer1/audit.json",
+        "/home/smlflg/Projekte/Meta2.0/data/layer1/session_digests.jsonl"
+      ],
+      "outputs": ["runs/<run_id>/manifests/baseline_snapshot.json"],
+      "entry_gate": "Meta2.0 artifacts exist.",
+      "exit_gate": "Coverage, unique digest count, raw row count, failures, and source spread are recorded."
+    },
+    {
+      "id": "p02_calibration_slice",
+      "kind": "small_model_run",
+      "purpose": "Answer a small representative slice before broad overnight execution.",
+      "inputs": ["baseline_snapshot.json", "config/questions.json", "config/experiment_design.json"],
+      "outputs": ["runs/<run_id>/outputs/calibration_answers.jsonl"],
+      "entry_gate": "Baseline audit is valid.",
+      "exit_gate": "At least one answer shows evidence, uncertainty, counterevidence status, and owner-ready conclusion."
+    },
+    {
+      "id": "p03_stratified_backfill",
+      "kind": "meta2_layer1_backfill",
+      "purpose": "Increase digest coverage while preserving source and time diversity.",
+      "inputs": ["baseline_snapshot.json"],
+      "outputs": ["updated Meta2.0 layer1 audit"],
+      "entry_gate": "Calibration did not reveal a method flaw.",
+      "exit_gate": "New audit shows no unresolved failures and improved coverage."
+    },
+    {
+      "id": "p04_claim_ledger_generation",
+      "kind": "structured_answer_run",
+      "purpose": "Generate structured claims per question instead of freeform essays.",
+      "inputs": ["session_digests.jsonl", "config/questions.json", "config/experiment_design.json"],
+      "outputs": ["runs/<run_id>/outputs/claim_ledger.jsonl"],
+      "entry_gate": "Coverage gate and requirements gate pass.",
+      "exit_gate": "Claims cite evidence and mark uncertainty."
+    },
+    {
+      "id": "p05_counterevidence_pass",
+      "kind": "adversarial_review",
+      "purpose": "Search for sessions that weaken, contradict, or narrow the generated claims.",
+      "inputs": ["claim_ledger.jsonl", "session_digests.jsonl"],
+      "outputs": ["runs/<run_id>/outputs/counterevidence_ledger.jsonl"],
+      "entry_gate": "Claim ledger exists.",
+      "exit_gate": "Each main claim has counterevidence status."
+    },
+    {
+      "id": "p06_validity_review",
+      "kind": "quality_gate",
+      "purpose": "Separate useful answers from invalid measurement.",
+      "inputs": ["claim_ledger.jsonl", "counterevidence_ledger.jsonl", "run_manifest.json", "batch_plan.json"],
+      "outputs": ["runs/<run_id>/reports/validity_review.md"],
+      "entry_gate": "Claim and counterevidence ledgers exist.",
+      "exit_gate": "Invalid, weak, and owner-ready outputs are explicitly separated."
+    },
+    {
+      "id": "p07_owner_synthesis",
+      "kind": "owner_report",
+      "purpose": "Compress validated claims into canonical owner-facing answers.",
+      "inputs": ["validity_review.md", "claim_ledger.jsonl", "counterevidence_ledger.jsonl"],
+      "outputs": ["runs/<run_id>/reports/OWNER_SYNTHESIS.md"],
+      "entry_gate": "Validity review passes.",
+      "exit_gate": "Each question has one thesis, evidence, limits, and next action."
+    },
+    {
+      "id": "p08_morning_closeout",
+      "kind": "closure",
+      "purpose": "Make the overnight run inspectable and stop hidden cycle debt.",
+      "inputs": ["run_manifest.json", "batch_plan.json", "logs", "reports"],
+      "outputs": ["runs/<run_id>/reports/MORNING_REPORT.md"],
+      "entry_gate": "Overnight run ended or was stopped.",
+      "exit_gate": "Actual calls, errors, outputs, missing evidence, and next decision are visible."
+    }
+  ]
+}

config/questions.json

@@ -0,0 +1,68 @@
+{
+  "version": 1,
+  "source": "GOAL",
+  "status": "draft_until_requirements_arrive",
+  "questions": [
+    {
+      "id": "q01_roter_faden",
+      "title": "Der rote Faden",
+      "prompt": "Welche 5 Themen kehren immer wieder zurueck, egal unter welchem Projektnamen?",
+      "answer_contract": ["pattern_list", "evidence_refs", "non_project_motifs", "uncertainty"]
+    },
+    {
+      "id": "q02_eigentliches_problem",
+      "title": "Das eigentliche Problem",
+      "prompt": "Welches Problem habe ich immer wieder neu formuliert, aber nie final benannt?",
+      "answer_contract": ["core_problem_hypothesis", "repeated_formulations", "counterexamples", "scope"]
+    },
+    {
+      "id": "q03_wiederkehrende_schleife",
+      "title": "Die wiederkehrende Schleife",
+      "prompt": "Welche Schleife wiederholt sich in meinen Sessions, und wo war Fortschritt statt rekursivem Kontrollverlust?",
+      "answer_contract": ["loop_steps", "progress_markers", "drift_markers", "break_conditions"]
+    },
+    {
+      "id": "q04_staerkste_beweise",
+      "title": "Die staerksten Beweise",
+      "prompt": "Welche Agenten-Sessions haben echte irreversible Evidenz in der Welt erzeugt?",
+      "answer_contract": ["world_contact_events", "artifacts", "people_or_money_or_deployments", "confidence"]
+    },
+    {
+      "id": "q05_selbsttaeuschungen",
+      "title": "Die Selbsttaeuschungen",
+      "prompt": "Wo habe ich Komplexitaet erzeugt, um Entscheidung, Angst oder Abschluss zu vermeiden?",
+      "answer_contract": ["avoidance_patterns", "architecture_vs_escape", "decision_debt", "evidence"]
+    },
+    {
+      "id": "q06_beste_arbeitsmodi",
+      "title": "Die besten Arbeitsmodi",
+      "prompt": "Unter welchen Bedingungen werde ich durch KI groesser, und unter welchen werde ich fragmentiert?",
+      "answer_contract": ["positive_conditions", "negative_conditions", "tools_models_context", "operating_rules"]
+    },
+    {
+      "id": "q07_ownership_test",
+      "title": "Der Ownership-Test",
+      "prompt": "Wann haben Agenten mir Macht gegeben, und wann haben sie mir mein mentales Modell geraubt?",
+      "answer_contract": ["ownership_gain", "ownership_loss", "explainability", "owner_gates"]
+    },
+    {
+      "id": "q08_produktlinie",
+      "title": "Die Produktlinie aus dem eigenen Leben",
+      "prompt": "Welche persoenlichen Probleme sind gleichzeitig Marktprobleme?",
+      "answer_contract": ["personal_problem", "market_problem", "evidence", "offer_shape"]
+    },
+    {
+      "id": "q09_unfertige_wahrheiten",
+      "title": "Die unfertigen Wahrheiten",
+      "prompt": "Welche Saetze tauchen immer wieder auf und sollten als Betriebssystem-Regeln operationalisiert werden?",
+      "answer_contract": ["recurring_sentences", "operational_rule", "where_to_enforce", "risk"]
+    },
+    {
+      "id": "q10_finale_verdichtung",
+      "title": "Die finale Verdichtung",
+      "prompt": "Was ist mein eigentliches Werk?",
+      "answer_contract": ["one_sentence_work", "evidence_tree", "not_this", "next_step"]
+    }
+  ]
+}
+

config/run_profile.example.json

@@ -0,0 +1,34 @@
+{
+  "version": 1,
+  "name": "meta3_overnight_profile_v0",
+  "mode": "prepare_only",
+  "source_artifacts": {
+    "meta2_root": "/home/smlflg/Projekte/Meta2.0",
+    "session_inventory": "/home/smlflg/Projekte/Meta2.0/data/session_inventory.json",
+    "layer1_digests": "/home/smlflg/Projekte/Meta2.0/data/layer1/session_digests.jsonl",
+    "layer1_audit": "/home/smlflg/Projekte/Meta2.0/data/layer1/audit.json"
+  },
+  "api_profile": {
+    "provider": "openai_compatible_litellm",
+    "default_base_url": "https://litellm-kommone.genai.govdigital.de/v1",
+    "base_url_envs": ["LITELLM_BASE_URL", "OPENAI_BASE_URL"],
+    "api_key_envs": ["LITELLM_API_KEYS", "OPENAI_API_KEY"],
+    "model_env": "META3_MODEL",
+    "default_model": "stackit-qwen-qwen3-vl-235b-a22b-instruct-fp8"
+  },
+  "limits": {
+    "max_api_calls": 10000000,
+    "safe_batch_size": 3,
+    "safe_concurrency": 2,
+    "max_concurrency_without_new_approval": 4,
+    "dry_run_required": true,
+    "network_calls_enabled_by_default": false
+  },
+  "planning": {
+    "layer1_backfill_enabled": true,
+    "answer_passes_per_question": 3,
+    "synthesis_passes": 2,
+    "minimum_layer1_coverage_pct_before_final_answers": 90.0
+  }
+}
+

data/.gitkeep

@@ -0,0 +1 @@
+

docs/META3_PROCESS.md

@@ -0,0 +1,94 @@
+# Meta3 Process
+
+Meta3.0 improves the meta process, not only the amount of compute.
+
+Meta2.0 answered: "Can we digest Samuel's session corpus at all?"
+
+Meta3.0 answers: "Which conclusions are valid enough for Samuel to use?"
+
+## Process Loop
+
+```text
+Requirements lock
+  -> baseline audit
+  -> calibration slice
+  -> stratified backfill
+  -> claim ledger
+  -> counterevidence pass
+  -> validity review
+  -> owner synthesis
+  -> morning closeout
+```
+
+## What Changes Against Meta2.0
+
+Meta2.0 strength:
+
+- stable inventory and digest pipeline
+- Qwen batch mechanics
+- audit files and coverage numbers
+
+Meta2.0 weakness for this task:
+
+- it is mostly a data-prep system
+- it does not decide which self-interpretations are valid
+- it does not force counterevidence before synthesis
+- it does not create an owner-ready answer contract by itself
+
+Meta3.0 improvement:
+
+- a run starts with requirements, not with fanout
+- every output belongs to a phase
+- every strong claim enters a ledger
+- every main claim receives a counterevidence pass
+- final synthesis is blocked until validity is reviewed
+- morning closeout records actual calls, errors, missing evidence, and the next
+  owner decision
+
+## Core Rule
+
+Do not scale a flawed loop.
+
+The first useful Meta3 run is therefore not the biggest possible run. It is the
+smallest loop that proves:
+
+1. Meta2.0 sources are measured correctly.
+2. The question set produces structured claims.
+3. Evidence references are concrete.
+4. Counterevidence is not skipped.
+5. The final answer gives Samuel one usable next mental model.
+
+## Required Artifacts Per Serious Run
+
+- `run_manifest.json`: what was attempted.
+- `batch_plan.json`: what was planned.
+- `baseline_snapshot.json`: what source state existed before scale-up.
+- `source_snapshot_manifest.json`: frozen local digest input and checksum.
+- `*_workpackets.jsonl`: exact prompt packets a future runner would consume.
+- `*_call_ledger.jsonl`: one planned ledger entry per workpacket/API call.
+- `calibration_answers.jsonl`: small proof before broad execution.
+- `claim_ledger.jsonl`: structured claims.
+- `counterevidence_ledger.jsonl`: contradictions, weak spots, limits.
+- `*_collection_report.json`: response parsing and schema-validation status.
+- `validity_review.md`: what is valid, weak, invalid, or missing.
+- `OWNER_SYNTHESIS.md`: owner-facing answers.
+- `MORNING_REPORT.md`: actual calls, errors, outputs, and next decision.
+
+## Failure Labels
+
+Use these labels instead of vague "worked/did not work":
+
+- `invalid_measurement`: artifacts missing, no comparable baseline, bad manifest.
+- `insufficient_coverage`: source coverage too low for final synthesis.
+- `weak_evidence`: claims exist but references are thin.
+- `counterevidence_missing`: claims were not attacked.
+- `owner_not_ready`: output is interesting but not usable by Samuel.
+- `ready_for_owner_review`: valid enough for Samuel to inspect.
+
+## Morning Question
+
+The next morning, do not ask "Did it run?"
+
+Ask:
+
+> Which claims survived evidence, counterevidence, and owner-readiness?

docs/OVERNIGHT_RUNBOOK.md

@@ -0,0 +1,237 @@
+# Overnight Runbook
+
+## Phase 0: Anforderungen aufnehmen
+
+1. `requirements/PENDING.md` ausfuellen.
+2. Nach Abstimmung `requirements/LOCKED.example.md` nach
+   `requirements/LOCKED.md` uebertragen und ausfuellen.
+3. `config/questions.json` anpassen oder ersetzen.
+4. `config/process_model.json` gegen die Anforderungen pruefen.
+5. Kein API-Start, solange diese Dateien nicht zusammenpassen.
+
+## Phase 1: Smoke
+
+```bash
+python3 scripts/meta3_smoke.py
+```
+
+Erwartung:
+
+- JSON valide.
+- Fragen-IDs eindeutig.
+- Meta2.0-Quellen existieren.
+- Meta3-Improvement-Vertrag existiert.
+- Scripts sind kompilierbar.
+- Keine Secrets in Config-Dateien.
+
+## Komplette lokale Vorbereitung
+
+Der lokale Orchestrator fuehrt die folgenden Phasen ohne Netzwerk nacheinander
+aus: Smoke, Prepare, Batchplan, Source-Freeze, Workpackets, Readiness,
+Runner-Dry-Runs, Runner-Status, Network-Preflight, Response-Collection und
+Morning-Report.
+
+```bash
+python3 scripts/meta3_orchestrate_local.py --accepted-call-budget 10000000
+```
+
+Erzeugt:
+
+```text
+runs/<run_id>/reports/local_orchestration_report.json
+runs/<run_id>/reports/LOCAL_ORCHESTRATION_REPORT.md
+```
+
+Solange `requirements/LOCKED.md` und API-Key-Env fehlen, ist der erwartete
+Status `prepared_blocked`, nicht `ready_for_network_start`.
+
+Wenn ein manueller Workpacket-Build mit `Source drift refused` stoppt, pruefe
+zuerst den Batchplan-Gate `layer1_audit_raw_digest_mismatch`. Dann sind
+Meta2.0-Audit und rohe `session_digests.jsonl` nicht synchron. Der lokale
+Orchestrator loest das durch `meta3_freeze_sources.py`, das einen reproduzierbar
+eingefrorenen Snapshot in den Run schreibt.
+
+## Phase 2: Run manifestieren
+
+```bash
+python3 scripts/meta3_prepare.py
+```
+
+Erzeugt:
+
+```text
+runs/<timestamp>_overnight_prepare/
+  manifests/run_manifest.json
+  state/
+  outputs/
+  logs/
+  tmp/
+```
+
+## Phase 3: Batchplan erzeugen
+
+```bash
+python3 scripts/meta3_plan_batches.py
+```
+
+Erzeugt:
+
+```text
+runs/<latest>/manifests/batch_plan.json
+```
+
+## Phase 3a: Source Snapshot einfrieren
+
+```bash
+python3 scripts/meta3_freeze_sources.py
+```
+
+Erzeugt:
+
+```text
+runs/<latest>/manifests/source_snapshot_manifest.json
+runs/<latest>/data/source_snapshot/layer1_digests.frozen.jsonl
+```
+
+Das ist die lokale stabile Quelle fuer Workpackets. Es macht keine API-Calls.
+
+## Phase 3b: Workpackets bauen
+
+Kalibrierung:
+
+```bash
+python3 scripts/meta3_build_workpackets.py --mode calibration --sample-size 24
+```
+
+Breiter Claim-Ledger-Plan gegen alle vorhandenen Digests:
+
+```bash
+python3 scripts/meta3_build_workpackets.py --mode claim_ledger --shard-size 24
+```
+
+Erzeugt:
+
+```text
+runs/<latest>/outputs/workpackets/<mode>_workpackets.jsonl
+runs/<latest>/outputs/workpackets/<mode>_packet_manifest.json
+runs/<latest>/state/<mode>_call_ledger.jsonl
+```
+
+Auch diese Phase macht keine API-Calls. Sie bereitet nur reproduzierbare Prompts
+und geplante Call-Ledger-Eintraege vor.
+
+## Phase 3c: Run-Artefakte validieren
+
+```bash
+python3 scripts/meta3_validate_run.py
+```
+
+Erzeugt:
+
+```text
+runs/<latest>/reports/run_readiness_report.json
+runs/<latest>/reports/RUN_READINESS_REPORT.md
+```
+
+Der Validator prueft Manifest, Batchplan, Baseline-Snapshot, Workpacket-Manifeste,
+Packet-Zeilen, Call-Ledger und die Null-Netzwerk-Invariante.
+
+## Phase 3d: Runner-Dry-Run
+
+```bash
+python3 scripts/meta3_run_packets.py --mode calibration --limit 1
+```
+
+Erzeugt:
+
+```text
+runs/<latest>/logs/calibration_runner_dry_run.json
+```
+
+Das prueft, welche Pakete ein Runner konsumieren wuerde, welche Concurrency
+gelten wuerde und ob Env-Namen gesetzt sind. Es macht keine API-Calls.
+
+Status/Heartbeat pruefen:
+
+```bash
+python3 scripts/meta3_runner_status.py
+```
+
+Stop-Dateien fuer echte Laeufe:
+
+```text
+runs/<run_id>/state/STOP
+runs/<run_id>/state/STOP_calibration
+runs/<run_id>/state/STOP_claim_ledger
+```
+
+Ein echter Start waere spaeter explizit, zum Beispiel:
+
+```bash
+python3 scripts/meta3_run_packets.py --mode calibration --limit 1 --execute-network --accepted-call-budget 1
+```
+
+Diesen Befehl nicht automatisch starten. Vorher braucht es Samuels Freigabe fuer
+Netzwerk, Kosten, Umfang und Concurrency.
+
+## Phase 3e: Netzwerk-Preflight
+
+```bash
+python3 scripts/meta3_preflight.py --mode calibration --accepted-call-budget 10
+```
+
+Erzeugt:
+
+```text
+runs/<latest>/reports/network_preflight.json
+runs/<latest>/reports/NETWORK_PREFLIGHT.md
+```
+
+Dieser Report ist der harte Startspiegel: Requirements-Lock, Readiness-Report,
+Runner-Dry-Run, API-Env, Budget und Concurrency muessen zusammenpassen. Ohne
+`requirements/LOCKED.md` bleibt der Status bewusst `blocked`.
+
+## Phase 4: Startentscheidung
+
+Vor einem echten Runner muessen sichtbar sein:
+
+- geplante API-Calls,
+- geplante Concurrency,
+- Quellenabdeckung,
+- wie Meta3 den Meta2-Baseline-Run methodisch verbessert,
+- welche Gegenbeweis- und Claim-Ledger-Pflichten gelten,
+- welche Workpackets ein Runner wirklich konsumieren wuerde,
+- offene Gates,
+- Abbruchbedingungen.
+
+## Phase 5: Morgen-Auswertung
+
+Responses einsammeln:
+
+```bash
+python3 scripts/meta3_collect_responses.py --mode calibration --allow-empty
+python3 scripts/meta3_collect_responses.py --mode claim_ledger --allow-empty
+```
+
+Morning-Report schreiben:
+
+```bash
+python3 scripts/meta3_morning_report.py
+```
+
+Morgen nicht nur "lief durch" pruefen, sondern:
+
+- wie viele Calls tatsaechlich,
+- wie viele Fehler,
+- wie viele leere/low-confidence Antworten,
+- welche Fragen beantwortet,
+- welche Evidenz fehlt,
+- welche Outputs owner-ready sind.
+
+Nullkosten-Test fuer die Parser-Strecke:
+
+```bash
+python3 scripts/meta3_create_mock_response.py --mode calibration
+python3 scripts/meta3_collect_responses.py --mode calibration
+python3 scripts/meta3_morning_report.py
+```

docs/SAFETY.md

@@ -0,0 +1,107 @@
+# Safety And Cost Gates
+
+## Keine Secrets
+
+Config-Dateien duerfen nur Env-Var-Namen enthalten, keine Werte.
+
+Erlaubt:
+
+```json
+"api_key_envs": ["LITELLM_API_KEYS", "OPENAI_API_KEY"]
+```
+
+Nicht erlaubt:
+
+```json
+"api_key": "..."
+```
+
+## Kein Start durch Scaffold
+
+Alle Scripts in `scripts/` sind lokal:
+
+- `meta3_smoke.py`
+- `meta3_prepare.py`
+- `meta3_plan_batches.py`
+- `meta3_build_workpackets.py`
+- `meta3_validate_run.py`
+- `meta3_preflight.py`
+
+Sie duerfen keine API-Requests ausfuehren.
+
+Ausnahme: `meta3_run_packets.py` kann spaeter Netzwerk nutzen, aber nur mit:
+
+- vorhandenen Workpackets,
+- vorherigem Dry-Run,
+- `--execute-network`,
+- `--accepted-call-budget <N>`,
+- vorhandener API-Key-Env.
+
+Ohne `--execute-network` schreibt es nur einen Dry-Run-Report.
+
+## Langlauf-Kontrolle
+
+Ein echter Runner schreibt und beachtet:
+
+- `state/<mode>_runner.lock.json`: verhindert Doppelstart.
+- `state/<mode>_runner_heartbeat.json`: Fortschritt, Fehler, In-Flight-Status.
+- `state/STOP`: stoppt alle Modi nach den laufenden In-Flight-Calls.
+- `state/STOP_<mode>`: stoppt nur einen Modus.
+
+Status pruefen:
+
+```bash
+python3 scripts/meta3_runner_status.py
+```
+
+Der Runner submitted nur bis zur Concurrency-Grenze gleichzeitig. Bei Stop- oder
+Fehlerschwelle werden keine neuen Pakete gestartet; laufende Pakete werden noch
+sauber verarbeitet.
+
+## Source-Drift-Gate
+
+Meta3 baut standardmaessig keine finalen Workpackets, wenn Meta2.0-Audit und
+rohe Digest-Datei auseinanderlaufen:
+
+```text
+layer1_audit_unique_digest_paths != layer1_raw_digest_file_rows_counted
+```
+
+Das verhindert, dass ein Overnight-Run gegen eine halb aktualisierte Quelle
+arbeitet. Der Batchplan markiert diesen Fall als `blocked`.
+`meta3_build_workpackets.py` verweigert ohne lokalen Source-Snapshot oder
+`--allow-source-drift`.
+
+Der sichere Pfad ist:
+
+```bash
+python3 scripts/meta3_freeze_sources.py
+```
+
+Danach konsumieren Workpackets die eingefrorene Datei unter
+`runs/<run_id>/data/source_snapshot/`. `--allow-source-drift` ist nur fuer
+diagnostische, nicht-finale Runs gedacht.
+
+## Call-Budget
+
+`max_api_calls=10000000` ist ein hartes Planungsmaximum, kein Startsignal.
+
+Vor echten Calls braucht es:
+
+- Run-Manifest,
+- Batchplan,
+- Workpacket-Manifest,
+- Call-Ledger,
+- Dry-Run-Report,
+- Network-Preflight-Report mit Status `pass`,
+- Kosten-/Concurrency-Akzeptanz,
+- Abbruchbedingungen.
+
+## Concurrency
+
+Start safe:
+
+- `safe_batch_size=3`
+- `safe_concurrency=2`
+
+Nicht direkt auf hohe Parallelitaet gehen. Erst Fehlerrate und 429-Rate sehen.

docs/evals/EVAL_meta3_improvement_over_meta2.md

@@ -0,0 +1,159 @@
+# Owner Packet
+
+## What This Eval Tests
+
+This eval checks whether Meta3.0 is an actual experimental improvement over
+Meta2.0, not just a new folder that reuses Meta2.0 batch mechanics.
+
+## Why It Matters
+
+The main failure risk is spending overnight Qwen budget on prettier synthesis
+without better validity: no stratified evidence, no counterevidence, no claim
+ledger, no owner-ready conclusion, and no proof that the run is comparable.
+
+## Release Meaning
+
+blocking
+
+Meta3.0 should not start a broad overnight run until this eval at least passes
+as an artifact/smoke check. Later it can become a runnable output-quality eval.
+
+## Smallest Next Implementation Step
+
+Keep `config/experiment_design.json` and this spec present, then enforce their
+required contract in `scripts/meta3_smoke.py`.
+
+# Eval Spec
+
+## 1. Eval Name
+
+Meta3 Improvement Over Meta2
+
+## 2. Purpose
+
+Protect against the regression where Meta3.0 becomes a Meta2.0 wrapper with
+more API calls but no stronger experiment design.
+
+## 3. Source
+
+Source type: expected product behavior and risk analysis.
+
+Samuel clarified that Meta3.0 must be "besser", "anders intelligenter" than
+Meta2.0 because it is the improvement. The risk is concrete because the first
+scaffold correctly reused Meta2.0 artifacts but did not yet make the improvement
+contract first-class.
+
+## 4. Eval Type
+
+human-ownership eval
+
+It checks both experiment validity and whether the resulting system keeps the
+owner in control of what counts as evidence, uncertainty, and final meaning.
+
+## 5. Scenario
+
+Before an overnight run, the owner asks whether Meta3.0 is ready to evaluate
+the ten self-analysis questions with a very large API budget. The system must
+prove that it has a better method than Meta2.0's data digest pipeline.
+
+## 6. Environment
+
+Required files:
+
+- `GOAL`
+- `config/questions.json`
+- `config/run_profile.example.json`
+- `config/experiment_design.json`
+- `ARCHITECTURE.md`
+- `docs/evals/EVAL_meta3_improvement_over_meta2.md`
+- `scripts/meta3_smoke.py`
+
+Required external artifacts, read-only:
+
+- `/home/smlflg/Projekte/Meta2.0/data/session_inventory.json`
+- `/home/smlflg/Projekte/Meta2.0/data/layer1/session_digests.jsonl`
+- `/home/smlflg/Projekte/Meta2.0/data/layer1/audit.json`
+
+No network access is required for this eval.
+
+## 7. Input
+
+Command:
+
+```bash
+python3 scripts/meta3_smoke.py
+```
+
+## 8. Expected Behavior
+
+The smoke check should pass only if Meta3.0 contains an explicit improvement
+contract with at least these capabilities:
+
+- stratified evidence
+- claim ledger
+- counterevidence pass
+- validity before synthesis
+- question graph
+- owner-ready output
+
+The docs should make clear that Meta2.0 remains the source/digest baseline, and
+Meta3.0 adds the experiment and synthesis controls.
+
+## 9. Failure Modes
+
+- `config/experiment_design.json` missing or invalid.
+- Required improvement requirement IDs missing.
+- The eval spec is missing.
+- The architecture does not distinguish Meta2.0 baseline role from Meta3.0
+  improvement role.
+- Smoke passes even though the improvement contract is absent.
+- Measurement failure: Meta2.0 artifacts are missing, but the run is described
+  as ready.
+
+## 10. Scoring
+
+Pass/fail.
+
+Pass means the artifact contract exists and the local smoke check validates the
+minimum improvement requirements. It does not yet prove answer quality.
+
+## 11. Assertions
+
+- `config/experiment_design.json` exists.
+- `schema` equals `meta3.experiment_design.v1`.
+- `improvement_requirements` contains:
+  - `stratified_evidence`
+  - `claim_ledger`
+  - `counterevidence_pass`
+  - `validity_before_synthesis`
+  - `question_graph`
+  - `owner_ready_output`
+- `docs/evals/EVAL_meta3_improvement_over_meta2.md` exists.
+- `ARCHITECTURE.md` contains `Meta3 Improvement Contract`.
+- No network calls are made.
+
+## 12. Minimal Runnable Form
+
+Evaluator logic lives in `scripts/meta3_smoke.py`.
+
+The minimal runnable command is:
+
+```bash
+python3 scripts/meta3_smoke.py
+```
+
+Expected result: `SMOKE PASS`, possibly with non-blocking warnings.
+
+## 13. Release Meaning
+
+blocking
+
+The check blocks broad overnight execution because otherwise Meta3.0 can spend
+large budget without proving that it improved on Meta2.0.
+
+## 14. Maintenance Notes
+
+Update this eval when Meta3.0 gains a real answer runner. At that point, add a
+second output-quality eval that inspects generated answers for evidence links,
+counterevidence, uncertainty, and owner-ready conclusions. Downgrade this eval
+only if a stronger runnable eval fully replaces it.

reports/.gitkeep

@@ -0,0 +1 @@
+

requirements/LOCKED.example.md

@@ -0,0 +1,45 @@
+# Meta3 Requirements Lock
+
+Status: locked
+Owner: Samuel
+Locked at: YYYY-MM-DDTHH:MM:SS+02:00
+
+## Ziel
+
+Welche konkrete Auswertung soll dieser Overnight-Run liefern?
+
+## Fragen
+
+Welche Fragen aus `config/questions.json` gelten fuer diesen Run?
+
+## Quellen
+
+Erlaubt:
+
+- `/home/smlflg/Projekte/Meta2.0/data/session_inventory.json`
+- `/home/smlflg/Projekte/Meta2.0/data/layer1/session_digests.jsonl`
+- `/home/smlflg/Projekte/Meta2.0/data/layer1/audit.json`
+
+Nicht erlaubt:
+
+- Secrets
+- globale Steuerflaechen ohne neue Freigabe
+
+## Budget
+
+- Max API calls:
+- Accepted initial call budget:
+- Max concurrency:
+- Max runtime:
+
+## Outputs
+
+Welche Dateien/Reports muessen morgen existieren?
+
+## Abbruchbedingungen
+
+Wann soll ein Runner stoppen?
+
+## Morgen-Erfolg
+
+Was zaehlt morgen als Erfolg, was nur als gelaufen?

requirements/PENDING.md

@@ -0,0 +1,33 @@
+# Requirements Intake
+
+Status: wartet auf Samuels konkrete Anforderungen.
+
+## Bitte hier eintragen
+
+```text
+Ziel:
+
+Welche Fragen:
+
+Welche Quellen duerfen gelesen werden:
+
+Welche Quellen duerfen NICHT gelesen werden:
+
+Maximale Kosten / Call-Budget:
+
+Maximale Laufzeit:
+
+Gewuenschte Outputs:
+
+Abbruchbedingungen:
+
+Was zaehlt morgen als Erfolg:
+```
+
+## Vorlaeufige Annahmen
+
+- Keine API-Calls aus diesem Ordner, bis Anforderungen da sind.
+- Externe Session-Stores bleiben read-only.
+- Meta2.0 bleibt Quelle fuer Inventory und Layer-1-Digests.
+- Der erste echte Run soll klein genug sein, dass er auditierbar bleibt.
+

runs/.gitkeep

@@ -0,0 +1 @@
+

runs/20260615T210314Z_overnight-prepare/manifests/batch_plan.json

@@ -0,0 +1,73 @@
+{
+  "schema": "meta3.batch_plan.v1",
+  "run_id": "20260615T210314Z_overnight-prepare",
+  "created_at": "2026-06-15T21:03:20.745357+00:00",
+  "network_calls_made": 0,
+  "source_snapshot": {
+    "inventory_total": 1104,
+    "layer1_processed": 96,
+    "layer1_coverage_pct": 8.7,
+    "layer1_digest_rows_counted": 135
+  },
+  "settings": {
+    "safe_batch_size": 3,
+    "safe_concurrency": 2,
+    "question_count": 10,
+    "answer_passes_per_question": 3,
+    "synthesis_passes": 2,
+    "max_api_calls": 10000000
+  },
+  "estimated_calls": {
+    "layer1_backfill": 336,
+    "question_answers": 30,
+    "synthesis": 2,
+    "total": 368
+  },
+  "phases": [
+    {
+      "id": "phase_0_requirements",
+      "kind": "manual_gate",
+      "status": "open",
+      "description": "Requirements in requirements/PENDING.md must be reconciled with config/questions.json."
+    },
+    {
+      "id": "phase_1_layer1_backfill",
+      "kind": "meta2_layer1",
+      "status": "planned",
+      "remaining_sessions": 1008,
+      "estimated_api_calls": 336,
+      "recommended_command": "cd /home/smlflg/Projekte/Meta2.0 && python3 scripts/meta2_layer1_digest.py --limit 120 --batch-size 3 --concurrency 2 --order source-balanced"
+    },
+    {
+      "id": "phase_2_question_answers",
+      "kind": "meta3_answers",
+      "status": "blocked_until_requirements_and_coverage",
+      "estimated_api_calls": 30
+    },
+    {
+      "id": "phase_3_synthesis",
+      "kind": "meta3_synthesis",
+      "status": "blocked_until_answers_exist",
+      "estimated_api_calls": 2
+    }
+  ],
+  "gates": [
+    {
+      "gate": "layer1_coverage_below_final_answer_threshold",
+      "status": "open",
+      "coverage_pct": 8.7,
+      "required_pct": 90.0,
+      "meaning": "Final answers should wait; layer1 backfill can still run."
+    },
+    {
+      "gate": "dry_run_required_before_network_calls",
+      "status": "open",
+      "meaning": "This scaffold did not run a network dry-run."
+    },
+    {
+      "gate": "network_calls_disabled_by_default",
+      "status": "open",
+      "meaning": "A future runner must require an explicit start flag."
+    }
+  ]
+}

runs/20260615T210314Z_overnight-prepare/manifests/run_manifest.json

@@ -0,0 +1,106 @@
+{
+  "schema": "meta3.run_manifest.v1",
+  "run_id": "20260615T210314Z_overnight-prepare",
+  "created_at": "2026-06-15T21:03:14.905378+00:00",
+  "root": "/home/smlflg/Projekte/Meta3.0",
+  "mode": "prepare_only",
+  "profile_path": "/home/smlflg/Projekte/Meta3.0/config/run_profile.example.json",
+  "questions_path": "/home/smlflg/Projekte/Meta3.0/config/questions.json",
+  "question_count": 10,
+  "question_ids": [
+    "q01_roter_faden",
+    "q02_eigentliches_problem",
+    "q03_wiederkehrende_schleife",
+    "q04_staerkste_beweise",
+    "q05_selbsttaeuschungen",
+    "q06_beste_arbeitsmodi",
+    "q07_ownership_test",
+    "q08_produktlinie",
+    "q09_unfertige_wahrheiten",
+    "q10_finale_verdichtung"
+  ],
+  "profile": {
+    "version": 1,
+    "name": "meta3_overnight_profile_v0",
+    "mode": "prepare_only",
+    "source_artifacts": {
+      "meta2_root": "/home/smlflg/Projekte/Meta2.0",
+      "session_inventory": "/home/smlflg/Projekte/Meta2.0/data/session_inventory.json",
+      "layer1_digests": "/home/smlflg/Projekte/Meta2.0/data/layer1/session_digests.jsonl",
+      "layer1_audit": "/home/smlflg/Projekte/Meta2.0/data/layer1/audit.json"
+    },
+    "api_profile": {
+      "provider": "openai_compatible_litellm",
+      "default_base_url": "https://litellm-kommone.genai.govdigital.de/v1",
+      "base_url_envs": [
+        "LITELLM_BASE_URL",
+        "OPENAI_BASE_URL"
+      ],
+      "api_key_envs": [
+        "LITELLM_API_KEYS",
+        "OPENAI_API_KEY"
+      ],
+      "model_env": "META3_MODEL",
+      "default_model": "stackit-qwen-qwen3-vl-235b-a22b-instruct-fp8"
+    },
+    "limits": {
+      "max_api_calls": 10000000,
+      "safe_batch_size": 3,
+      "safe_concurrency": 2,
+      "max_concurrency_without_new_approval": 4,
+      "dry_run_required": true,
+      "network_calls_enabled_by_default": false
+    },
+    "planning": {
+      "layer1_backfill_enabled": true,
+      "answer_passes_per_question": 3,
+      "synthesis_passes": 2,
+      "minimum_layer1_coverage_pct_before_final_answers": 90.0
+    }
+  },
+  "questions_version": 1,
+  "source_status": {
+    "meta2_root": {
+      "path": "/home/smlflg/Projekte/Meta2.0",
+      "exists": true,
+      "is_file": false,
+      "is_dir": true,
+      "size_bytes": null
+    },
+    "session_inventory": {
+      "path": "/home/smlflg/Projekte/Meta2.0/data/session_inventory.json",
+      "exists": true,
+      "is_file": true,
+      "is_dir": false,
+      "size_bytes": 615014
+    },
+    "layer1_digests": {
+      "path": "/home/smlflg/Projekte/Meta2.0/data/layer1/session_digests.jsonl",
+      "exists": true,
+      "is_file": true,
+      "is_dir": false,
+      "size_bytes": 175726
+    },
+    "layer1_audit": {
+      "path": "/home/smlflg/Projekte/Meta2.0/data/layer1/audit.json",
+      "exists": true,
+      "is_file": true,
+      "is_dir": false,
+      "size_bytes": 763
+    }
+  },
+  "env_status": {
+    "api_key_envs_present": {
+      "LITELLM_API_KEYS": false,
+      "OPENAI_API_KEY": false
+    },
+    "base_url_envs_present": {
+      "LITELLM_BASE_URL": false,
+      "OPENAI_BASE_URL": false
+    },
+    "model_env_present": false,
+    "secrets_recorded": false
+  },
+  "network_calls_made": 0,
+  "next_step": "Run scripts/meta3_plan_batches.py --run-id 20260615T210314Z_overnight-prepare"
+}

runs/20260615T210908Z_improved-process-smoke/manifests/baseline_snapshot.json

@@ -0,0 +1,43 @@
+{
+  "schema": "meta3.baseline_snapshot.v1",
+  "run_id": "20260615T210908Z_improved-process-smoke",
+  "created_at": "2026-06-15T21:10:49.504026+00:00",
+  "baseline_project": "/home/smlflg/Projekte/Meta2.0",
+  "source_artifacts": {
+    "meta2_root": "/home/smlflg/Projekte/Meta2.0",
+    "session_inventory": "/home/smlflg/Projekte/Meta2.0/data/session_inventory.json",
+    "layer1_digests": "/home/smlflg/Projekte/Meta2.0/data/layer1/session_digests.jsonl",
+    "layer1_audit": "/home/smlflg/Projekte/Meta2.0/data/layer1/audit.json"
+  },
+  "source_snapshot": {
+    "inventory_total": 1104,
+    "layer1_processed": 216,
+    "layer1_coverage_pct": 19.57,
+    "layer1_audit_digest_rows": 216,
+    "layer1_audit_unique_digest_paths": 216,
+    "layer1_raw_digest_file_rows_counted": 216,
+    "unresolved_failures": 0,
+    "raw_failure_rows": 6,
+    "by_source": {
+      "claude_projects": 109,
+      "codex_sessions": 103,
+      "hermes_tirith_logs": 3,
+      "hermes_codex_sessions": 1
+    },
+    "confidence": {
+      "high": 156,
+      "medium": 48,
+      "low": 12
+    },
+    "strategic_relevance": {
+      "infrastructure": 151,
+      "projects": 133,
+      "hai": 63,
+      "harness": 19,
+      "overload": 17,
+      "monetization": 11,
+      "security": 2
+    }
+  },
+  "network_calls_made": 0
+}

runs/20260615T210908Z_improved-process-smoke/manifests/batch_plan.json

@@ -0,0 +1,151 @@
+{
+  "schema": "meta3.batch_plan.v1",
+  "run_id": "20260615T210908Z_improved-process-smoke",
+  "created_at": "2026-06-15T21:10:49.504295+00:00",
+  "network_calls_made": 0,
+  "source_snapshot": {
+    "inventory_total": 1104,
+    "layer1_processed": 216,
+    "layer1_coverage_pct": 19.57,
+    "layer1_audit_digest_rows": 216,
+    "layer1_audit_unique_digest_paths": 216,
+    "layer1_raw_digest_file_rows_counted": 216,
+    "unresolved_failures": 0,
+    "raw_failure_rows": 6,
+    "by_source": {
+      "claude_projects": 109,
+      "codex_sessions": 103,
+      "hermes_tirith_logs": 3,
+      "hermes_codex_sessions": 1
+    },
+    "confidence": {
+      "high": 156,
+      "medium": 48,
+      "low": 12
+    },
+    "strategic_relevance": {
+      "infrastructure": 151,
+      "projects": 133,
+      "hai": 63,
+      "harness": 19,
+      "overload": 17,
+      "monetization": 11,
+      "security": 2
+    }
+  },
+  "baseline_snapshot_path": "/home/smlflg/Projekte/Meta3.0/runs/20260615T210908Z_improved-process-smoke/manifests/baseline_snapshot.json",
+  "settings": {
+    "safe_batch_size": 3,
+    "safe_concurrency": 2,
+    "question_count": 10,
+    "answer_passes_per_question": 3,
+    "synthesis_passes": 2,
+    "max_api_calls": 10000000
+  },
+  "process_model": {
+    "phase_count": 9,
+    "source": "/home/smlflg/Projekte/Meta3.0/config/process_model.json"
+  },
+  "estimated_calls": {
+    "layer1_backfill": 296,
+    "claim_ledger_generation": 30,
+    "owner_synthesis": 2,
+    "total": 328
+  },
+  "phases": [
+    {
+      "id": "p00_requirements_lock",
+      "kind": "manual_gate",
+      "status": "open",
+      "purpose": "Lock the owner question, allowed sources, stop conditions, budget, and morning success definition.",
+      "exit_gate": "Questions, constraints, and success criteria are reconciled.",
+      "blocking_reason": "requirements/PENDING.md has not been locked."
+    },
+    {
+      "id": "p01_baseline_audit",
+      "kind": "measurement_gate",
+      "status": "complete",
+      "purpose": "Measure the Meta2.0 baseline before changing scale.",
+      "exit_gate": "Coverage, unique digest count, raw row count, failures, and source spread are recorded.",
+      "inventory_total": 1104,
+      "coverage_pct": 19.57
+    },
+    {
+      "id": "p02_calibration_slice",
+      "kind": "small_model_run",
+      "status": "blocked_until_requirements_lock",
+      "purpose": "Answer a small representative slice before broad overnight execution.",
+      "exit_gate": "At least one answer shows evidence, uncertainty, counterevidence status, and owner-ready conclusion.",
+      "estimated_api_calls": 0
+    },
+    {
+      "id": "p03_stratified_backfill",
+      "kind": "meta2_layer1_backfill",
+      "status": "planned",
+      "purpose": "Increase digest coverage while preserving source and time diversity.",
+      "exit_gate": "New audit shows no unresolved failures and improved coverage.",
+      "remaining_sessions": 888,
+      "estimated_api_calls": 296,
+      "recommended_command": "cd /home/smlflg/Projekte/Meta2.0 && python3 scripts/meta2_layer1_digest.py --limit 120 --batch-size 3 --concurrency 2 --order source-balanced"
+    },
+    {
+      "id": "p04_claim_ledger_generation",
+      "kind": "structured_answer_run",
+      "status": "blocked_until_requirements_calibration_and_coverage",
+      "purpose": "Generate structured claims per question instead of freeform essays.",
+      "exit_gate": "Claims cite evidence and mark uncertainty.",
+      "estimated_api_calls": 30
+    },
+    {
+      "id": "p05_counterevidence_pass",
+      "kind": "adversarial_review",
+      "status": "blocked_until_claim_ledger",
+      "purpose": "Search for sessions that weaken, contradict, or narrow the generated claims.",
+      "exit_gate": "Each main claim has counterevidence status.",
+      "estimated_api_calls": 0
+    },
+    {
+      "id": "p06_validity_review",
+      "kind": "quality_gate",
+      "status": "blocked_until_claim_and_counterevidence_ledgers",
+      "purpose": "Separate useful answers from invalid measurement.",
+      "exit_gate": "Invalid, weak, and owner-ready outputs are explicitly separated.",
+      "estimated_api_calls": 0
+    },
+    {
+      "id": "p07_owner_synthesis",
+      "kind": "owner_report",
+      "status": "blocked_until_validity_review",
+      "purpose": "Compress validated claims into canonical owner-facing answers.",
+      "exit_gate": "Each question has one thesis, evidence, limits, and next action.",
+      "estimated_api_calls": 2
+    },
+    {
+      "id": "p08_morning_closeout",
+      "kind": "closure",
+      "status": "planned_after_run_end",
+      "purpose": "Make the overnight run inspectable and stop hidden cycle debt.",
+      "exit_gate": "Actual calls, errors, outputs, missing evidence, and next decision are visible.",
+      "estimated_api_calls": 0
+    }
+  ],
+  "gates": [
+    {
+      "gate": "layer1_coverage_below_final_answer_threshold",
+      "status": "open",
+      "coverage_pct": 19.57,
+      "required_pct": 90.0,
+      "meaning": "Final answers should wait; layer1 backfill can still run."
+    },
+    {
+      "gate": "dry_run_required_before_network_calls",
+      "status": "open",
+      "meaning": "This scaffold did not run a network dry-run."
+    },
+    {
+      "gate": "network_calls_disabled_by_default",
+      "status": "open",
+      "meaning": "A future runner must require an explicit start flag."
+    }
+  ]
+}

runs/20260615T210908Z_improved-process-smoke/manifests/run_manifest.json

@@ -0,0 +1,122 @@
+{
+  "schema": "meta3.run_manifest.v1",
+  "run_id": "20260615T210908Z_improved-process-smoke",
+  "created_at": "2026-06-15T21:09:08.358930+00:00",
+  "root": "/home/smlflg/Projekte/Meta3.0",
+  "mode": "prepare_only",
+  "profile_path": "/home/smlflg/Projekte/Meta3.0/config/run_profile.example.json",
+  "questions_path": "/home/smlflg/Projekte/Meta3.0/config/questions.json",
+  "process_model_path": "/home/smlflg/Projekte/Meta3.0/config/process_model.json",
+  "experiment_design_path": "/home/smlflg/Projekte/Meta3.0/config/experiment_design.json",
+  "question_count": 10,
+  "question_ids": [
+    "q01_roter_faden",
+    "q02_eigentliches_problem",
+    "q03_wiederkehrende_schleife",
+    "q04_staerkste_beweise",
+    "q05_selbsttaeuschungen",
+    "q06_beste_arbeitsmodi",
+    "q07_ownership_test",
+    "q08_produktlinie",
+    "q09_unfertige_wahrheiten",
+    "q10_finale_verdichtung"
+  ],
+  "process_phase_count": 9,
+  "process_phase_ids": [
+    "p00_requirements_lock",
+    "p01_baseline_audit",
+    "p02_calibration_slice",
+    "p03_stratified_backfill",
+    "p04_claim_ledger_generation",
+    "p05_counterevidence_pass",
+    "p06_validity_review",
+    "p07_owner_synthesis",
+    "p08_morning_closeout"
+  ],
+  "profile": {
+    "version": 1,
+    "name": "meta3_overnight_profile_v0",
+    "mode": "prepare_only",
+    "source_artifacts": {
+      "meta2_root": "/home/smlflg/Projekte/Meta2.0",
+      "session_inventory": "/home/smlflg/Projekte/Meta2.0/data/session_inventory.json",
+      "layer1_digests": "/home/smlflg/Projekte/Meta2.0/data/layer1/session_digests.jsonl",
+      "layer1_audit": "/home/smlflg/Projekte/Meta2.0/data/layer1/audit.json"
+    },
+    "api_profile": {
+      "provider": "openai_compatible_litellm",
+      "default_base_url": "https://litellm-kommone.genai.govdigital.de/v1",
+      "base_url_envs": [
+        "LITELLM_BASE_URL",
+        "OPENAI_BASE_URL"
+      ],
+      "api_key_envs": [
+        "LITELLM_API_KEYS",
+        "OPENAI_API_KEY"
+      ],
+      "model_env": "META3_MODEL",
+      "default_model": "stackit-qwen-qwen3-vl-235b-a22b-instruct-fp8"
+    },
+    "limits": {
+      "max_api_calls": 10000000,
+      "safe_batch_size": 3,
+      "safe_concurrency": 2,
+      "max_concurrency_without_new_approval": 4,
+      "dry_run_required": true,
+      "network_calls_enabled_by_default": false
+    },
+    "planning": {
+      "layer1_backfill_enabled": true,
+      "answer_passes_per_question": 3,
+      "synthesis_passes": 2,
+      "minimum_layer1_coverage_pct_before_final_answers": 90.0
+    }
+  },
+  "process_model_schema": "meta3.process_model.v1",
+  "experiment_design_schema": "meta3.experiment_design.v1",
+  "questions_version": 1,
+  "source_status": {
+    "meta2_root": {
+      "path": "/home/smlflg/Projekte/Meta2.0",
+      "exists": true,
+      "is_file": false,
+      "is_dir": true,
+      "size_bytes": null
+    },
+    "session_inventory": {
+      "path": "/home/smlflg/Projekte/Meta2.0/data/session_inventory.json",
+      "exists": true,
+      "is_file": true,
+      "is_dir": false,
+      "size_bytes": 615014
+    },
+    "layer1_digests": {
+      "path": "/home/smlflg/Projekte/Meta2.0/data/layer1/session_digests.jsonl",
+      "exists": true,
+      "is_file": true,
+      "is_dir": false,
+      "size_bytes": 298916
+    },
+    "layer1_audit": {
+      "path": "/home/smlflg/Projekte/Meta2.0/data/layer1/audit.json",
+      "exists": true,
+      "is_file": true,
+      "is_dir": false,
+      "size_bytes": 795
+    }
+  },
+  "env_status": {
+    "api_key_envs_present": {
+      "LITELLM_API_KEYS": false,
+      "OPENAI_API_KEY": false
+    },
+    "base_url_envs_present": {
+      "LITELLM_BASE_URL": false,
+      "OPENAI_BASE_URL": false
+    },
+    "model_env_present": false,
+    "secrets_recorded": false
+  },
+  "network_calls_made": 0,
+  "next_step": "Run scripts/meta3_plan_batches.py --run-id 20260615T210908Z_improved-process-smoke"
+}

runs/20260615T210908Z_improved-process-smoke/outputs/workpackets/calibration_packet_manifest.json

@@ -0,0 +1,25 @@
+{
+  "schema": "meta3.workpacket_manifest.v1",
+  "run_id": "20260615T210908Z_improved-process-smoke",
+  "mode": "calibration",
+  "created_at": "2026-06-15T21:14:32.014347+00:00",
+  "network_calls_made": 0,
+  "source_digest_path": "/home/smlflg/Projekte/Meta2.0/data/layer1/session_digests.jsonl",
+  "available_digest_rows": 216,
+  "packet_count": 10,
+  "planned_call_count": 10,
+  "sample_size": 24,
+  "shard_size": 24,
+  "question_count": 10,
+  "unique_digest_refs_in_packets": 24,
+  "source_family_mentions": {
+    "claude_projects": 100,
+    "codex_sessions": 100,
+    "hermes_codex_sessions": 10,
+    "hermes_tirith_logs": 30
+  },
+  "packet_path": "/home/smlflg/Projekte/Meta3.0/runs/20260615T210908Z_improved-process-smoke/outputs/workpackets/calibration_workpackets.jsonl",
+  "call_ledger_path": "/home/smlflg/Projekte/Meta3.0/runs/20260615T210908Z_improved-process-smoke/state/calibration_call_ledger.jsonl",
+  "expected_response": "strict JSON per packet.expected_output_contract",
+  "next_step": "A future network runner may consume this packet file only after explicit approval."
+}

runs/20260615T210908Z_improved-process-smoke/outputs/workpackets/claim_ledger_packet_manifest.json

@@ -0,0 +1,26 @@
+{
+  "schema": "meta3.workpacket_manifest.v1",
+  "run_id": "20260615T210908Z_improved-process-smoke",
+  "mode": "claim_ledger",
+  "created_at": "2026-06-15T21:14:32.204479+00:00",
+  "network_calls_made": 0,
+  "source_digest_path": "/home/smlflg/Projekte/Meta2.0/data/layer1/session_digests.jsonl",
+  "available_digest_rows": 216,
+  "packet_count": 90,
+  "planned_call_count": 90,
+  "sample_size": 24,
+  "shard_size": 24,
+  "question_count": 10,
+  "unique_digest_refs_in_packets": 216,
+  "source_family_mentions": {
+    "claude_projects": 1090,
+    "codex_sessions": 1030,
+    "hermes_codex_sessions": 10,
+    "hermes_tirith_logs": 30
+  },
+  "packet_path": "/home/smlflg/Projekte/Meta3.0/runs/20260615T210908Z_improved-process-smoke/outputs/workpackets/claim_ledger_workpackets.jsonl",
+  "call_ledger_path": "/home/smlflg/Projekte/Meta3.0/runs/20260615T210908Z_improved-process-smoke/state/claim_ledger_call_ledger.jsonl",
+  "expected_response": "strict JSON per packet.expected_output_contract",
+  "next_step": "A future network runner may consume this packet file only after explicit approval.",
+  "shard_count_per_question": 9
+}

runs/20260615T210908Z_improved-process-smoke/reports/RUN_READINESS_REPORT.md

@@ -0,0 +1,16 @@
+# Meta3 Run Readiness Report
+
+Run: `20260615T210908Z_improved-process-smoke`
+Created: `2026-06-15T21:15:50.290399+00:00`
+Status: `pass`
+
+## Counts
+
+- Errors: 0
+- Warnings: 0
+- calibration: 10 packets, 10 planned calls, 24 unique digest refs
+- claim_ledger: 90 packets, 90 planned calls, 216 unique digest refs
+
+## Network
+
+- network_calls_made: 0

runs/20260615T210908Z_improved-process-smoke/reports/run_readiness_report.json

@@ -0,0 +1,37 @@
+{
+  "schema": "meta3.run_readiness_report.v1",
+  "run_id": "20260615T210908Z_improved-process-smoke",
+  "created_at": "2026-06-15T21:15:50.290399+00:00",
+  "status": "pass",
+  "required_modes": [
+    "calibration",
+    "claim_ledger"
+  ],
+  "workpacket_modes": {
+    "calibration": {
+      "packet_count": 10,
+      "planned_call_count": 10,
+      "unique_digest_refs_in_packets": 24,
+      "source_family_mentions": {
+        "claude_projects": 100,
+        "codex_sessions": 100,
+        "hermes_codex_sessions": 10,
+        "hermes_tirith_logs": 30
+      }
+    },
+    "claim_ledger": {
+      "packet_count": 90,
+      "planned_call_count": 90,
+      "unique_digest_refs_in_packets": 216,
+      "source_family_mentions": {
+        "claude_projects": 1090,
+        "codex_sessions": 1030,
+        "hermes_codex_sessions": 10,
+        "hermes_tirith_logs": 30
+      }
+    }
+  },
+  "errors": [],
+  "warnings": [],
+  "network_calls_made": 0
+}

runs/20260615T210908Z_improved-process-smoke/state/calibration_call_ledger.jsonl

@@ -0,0 +1,10 @@
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.005044+00:00", "error": null, "mode": "calibration", "network_calls_made": 0, "output_path": null, "packet_id": "calibration__q01_roter_faden__shard-0001", "question_id": "q01_roter_faden", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.005044+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.005044+00:00", "error": null, "mode": "calibration", "network_calls_made": 0, "output_path": null, "packet_id": "calibration__q02_eigentliches_problem__shard-0001", "question_id": "q02_eigentliches_problem", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.005044+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.005044+00:00", "error": null, "mode": "calibration", "network_calls_made": 0, "output_path": null, "packet_id": "calibration__q03_wiederkehrende_schleife__shard-0001", "question_id": "q03_wiederkehrende_schleife", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.005044+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.005044+00:00", "error": null, "mode": "calibration", "network_calls_made": 0, "output_path": null, "packet_id": "calibration__q04_staerkste_beweise__shard-0001", "question_id": "q04_staerkste_beweise", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.005044+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.005044+00:00", "error": null, "mode": "calibration", "network_calls_made": 0, "output_path": null, "packet_id": "calibration__q05_selbsttaeuschungen__shard-0001", "question_id": "q05_selbsttaeuschungen", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.005044+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.005044+00:00", "error": null, "mode": "calibration", "network_calls_made": 0, "output_path": null, "packet_id": "calibration__q06_beste_arbeitsmodi__shard-0001", "question_id": "q06_beste_arbeitsmodi", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.005044+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.005044+00:00", "error": null, "mode": "calibration", "network_calls_made": 0, "output_path": null, "packet_id": "calibration__q07_ownership_test__shard-0001", "question_id": "q07_ownership_test", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.005044+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.005044+00:00", "error": null, "mode": "calibration", "network_calls_made": 0, "output_path": null, "packet_id": "calibration__q08_produktlinie__shard-0001", "question_id": "q08_produktlinie", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.005044+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.005044+00:00", "error": null, "mode": "calibration", "network_calls_made": 0, "output_path": null, "packet_id": "calibration__q09_unfertige_wahrheiten__shard-0001", "question_id": "q09_unfertige_wahrheiten", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.005044+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.005044+00:00", "error": null, "mode": "calibration", "network_calls_made": 0, "output_path": null, "packet_id": "calibration__q10_finale_verdichtung__shard-0001", "question_id": "q10_finale_verdichtung", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.005044+00:00"}

runs/20260615T210908Z_improved-process-smoke/state/claim_ledger_call_ledger.jsonl

@@ -0,0 +1,90 @@
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q01_roter_faden__shard-0001", "question_id": "q01_roter_faden", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q01_roter_faden__shard-0002", "question_id": "q01_roter_faden", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q01_roter_faden__shard-0003", "question_id": "q01_roter_faden", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q01_roter_faden__shard-0004", "question_id": "q01_roter_faden", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q01_roter_faden__shard-0005", "question_id": "q01_roter_faden", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q01_roter_faden__shard-0006", "question_id": "q01_roter_faden", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q01_roter_faden__shard-0007", "question_id": "q01_roter_faden", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q01_roter_faden__shard-0008", "question_id": "q01_roter_faden", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q01_roter_faden__shard-0009", "question_id": "q01_roter_faden", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q02_eigentliches_problem__shard-0001", "question_id": "q02_eigentliches_problem", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q02_eigentliches_problem__shard-0002", "question_id": "q02_eigentliches_problem", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q02_eigentliches_problem__shard-0003", "question_id": "q02_eigentliches_problem", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q02_eigentliches_problem__shard-0004", "question_id": "q02_eigentliches_problem", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q02_eigentliches_problem__shard-0005", "question_id": "q02_eigentliches_problem", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q02_eigentliches_problem__shard-0006", "question_id": "q02_eigentliches_problem", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q02_eigentliches_problem__shard-0007", "question_id": "q02_eigentliches_problem", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q02_eigentliches_problem__shard-0008", "question_id": "q02_eigentliches_problem", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q02_eigentliches_problem__shard-0009", "question_id": "q02_eigentliches_problem", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q03_wiederkehrende_schleife__shard-0001", "question_id": "q03_wiederkehrende_schleife", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q03_wiederkehrende_schleife__shard-0002", "question_id": "q03_wiederkehrende_schleife", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q03_wiederkehrende_schleife__shard-0003", "question_id": "q03_wiederkehrende_schleife", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q03_wiederkehrende_schleife__shard-0004", "question_id": "q03_wiederkehrende_schleife", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q03_wiederkehrende_schleife__shard-0005", "question_id": "q03_wiederkehrende_schleife", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q03_wiederkehrende_schleife__shard-0006", "question_id": "q03_wiederkehrende_schleife", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q03_wiederkehrende_schleife__shard-0007", "question_id": "q03_wiederkehrende_schleife", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q03_wiederkehrende_schleife__shard-0008", "question_id": "q03_wiederkehrende_schleife", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q03_wiederkehrende_schleife__shard-0009", "question_id": "q03_wiederkehrende_schleife", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q04_staerkste_beweise__shard-0001", "question_id": "q04_staerkste_beweise", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q04_staerkste_beweise__shard-0002", "question_id": "q04_staerkste_beweise", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q04_staerkste_beweise__shard-0003", "question_id": "q04_staerkste_beweise", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q04_staerkste_beweise__shard-0004", "question_id": "q04_staerkste_beweise", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q04_staerkste_beweise__shard-0005", "question_id": "q04_staerkste_beweise", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q04_staerkste_beweise__shard-0006", "question_id": "q04_staerkste_beweise", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q04_staerkste_beweise__shard-0007", "question_id": "q04_staerkste_beweise", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q04_staerkste_beweise__shard-0008", "question_id": "q04_staerkste_beweise", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q04_staerkste_beweise__shard-0009", "question_id": "q04_staerkste_beweise", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q05_selbsttaeuschungen__shard-0001", "question_id": "q05_selbsttaeuschungen", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q05_selbsttaeuschungen__shard-0002", "question_id": "q05_selbsttaeuschungen", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q05_selbsttaeuschungen__shard-0003", "question_id": "q05_selbsttaeuschungen", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q05_selbsttaeuschungen__shard-0004", "question_id": "q05_selbsttaeuschungen", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q05_selbsttaeuschungen__shard-0005", "question_id": "q05_selbsttaeuschungen", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q05_selbsttaeuschungen__shard-0006", "question_id": "q05_selbsttaeuschungen", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q05_selbsttaeuschungen__shard-0007", "question_id": "q05_selbsttaeuschungen", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q05_selbsttaeuschungen__shard-0008", "question_id": "q05_selbsttaeuschungen", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q05_selbsttaeuschungen__shard-0009", "question_id": "q05_selbsttaeuschungen", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q06_beste_arbeitsmodi__shard-0001", "question_id": "q06_beste_arbeitsmodi", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q06_beste_arbeitsmodi__shard-0002", "question_id": "q06_beste_arbeitsmodi", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q06_beste_arbeitsmodi__shard-0003", "question_id": "q06_beste_arbeitsmodi", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q06_beste_arbeitsmodi__shard-0004", "question_id": "q06_beste_arbeitsmodi", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q06_beste_arbeitsmodi__shard-0005", "question_id": "q06_beste_arbeitsmodi", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q06_beste_arbeitsmodi__shard-0006", "question_id": "q06_beste_arbeitsmodi", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q06_beste_arbeitsmodi__shard-0007", "question_id": "q06_beste_arbeitsmodi", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q06_beste_arbeitsmodi__shard-0008", "question_id": "q06_beste_arbeitsmodi", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q06_beste_arbeitsmodi__shard-0009", "question_id": "q06_beste_arbeitsmodi", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q07_ownership_test__shard-0001", "question_id": "q07_ownership_test", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q07_ownership_test__shard-0002", "question_id": "q07_ownership_test", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q07_ownership_test__shard-0003", "question_id": "q07_ownership_test", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q07_ownership_test__shard-0004", "question_id": "q07_ownership_test", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q07_ownership_test__shard-0005", "question_id": "q07_ownership_test", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q07_ownership_test__shard-0006", "question_id": "q07_ownership_test", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q07_ownership_test__shard-0007", "question_id": "q07_ownership_test", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q07_ownership_test__shard-0008", "question_id": "q07_ownership_test", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q07_ownership_test__shard-0009", "question_id": "q07_ownership_test", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q08_produktlinie__shard-0001", "question_id": "q08_produktlinie", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q08_produktlinie__shard-0002", "question_id": "q08_produktlinie", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q08_produktlinie__shard-0003", "question_id": "q08_produktlinie", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q08_produktlinie__shard-0004", "question_id": "q08_produktlinie", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q08_produktlinie__shard-0005", "question_id": "q08_produktlinie", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q08_produktlinie__shard-0006", "question_id": "q08_produktlinie", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q08_produktlinie__shard-0007", "question_id": "q08_produktlinie", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q08_produktlinie__shard-0008", "question_id": "q08_produktlinie", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q08_produktlinie__shard-0009", "question_id": "q08_produktlinie", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q09_unfertige_wahrheiten__shard-0001", "question_id": "q09_unfertige_wahrheiten", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q09_unfertige_wahrheiten__shard-0002", "question_id": "q09_unfertige_wahrheiten", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q09_unfertige_wahrheiten__shard-0003", "question_id": "q09_unfertige_wahrheiten", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q09_unfertige_wahrheiten__shard-0004", "question_id": "q09_unfertige_wahrheiten", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q09_unfertige_wahrheiten__shard-0005", "question_id": "q09_unfertige_wahrheiten", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q09_unfertige_wahrheiten__shard-0006", "question_id": "q09_unfertige_wahrheiten", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q09_unfertige_wahrheiten__shard-0007", "question_id": "q09_unfertige_wahrheiten", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q09_unfertige_wahrheiten__shard-0008", "question_id": "q09_unfertige_wahrheiten", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q09_unfertige_wahrheiten__shard-0009", "question_id": "q09_unfertige_wahrheiten", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q10_finale_verdichtung__shard-0001", "question_id": "q10_finale_verdichtung", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q10_finale_verdichtung__shard-0002", "question_id": "q10_finale_verdichtung", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q10_finale_verdichtung__shard-0003", "question_id": "q10_finale_verdichtung", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q10_finale_verdichtung__shard-0004", "question_id": "q10_finale_verdichtung", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q10_finale_verdichtung__shard-0005", "question_id": "q10_finale_verdichtung", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q10_finale_verdichtung__shard-0006", "question_id": "q10_finale_verdichtung", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q10_finale_verdichtung__shard-0007", "question_id": "q10_finale_verdichtung", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q10_finale_verdichtung__shard-0008", "question_id": "q10_finale_verdichtung", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}
+{"attempts": 0, "created_at": "2026-06-15T21:14:32.129801+00:00", "error": null, "mode": "claim_ledger", "network_calls_made": 0, "output_path": null, "packet_id": "claim_ledger__q10_finale_verdichtung__shard-0009", "question_id": "q10_finale_verdichtung", "run_id": "20260615T210908Z_improved-process-smoke", "schema": "meta3.call_ledger_entry.v1", "status": "planned", "updated_at": "2026-06-15T21:14:32.129801+00:00"}

runs/20260615T211556Z_overnight-ready-smoke/logs/calibration_runner_dry_run.json

@@ -0,0 +1,15 @@
+{
+  "schema": "meta3.runner_dry_run.v1",
+  "run_id": "20260615T211556Z_overnight-ready-smoke",
+  "mode": "calibration",
+  "created_at": "2026-06-15T21:18:25.729188+00:00",
+  "execute_network": false,
+  "selected_packet_count": 1,
+  "concurrency": 2,
+  "model": "stackit-qwen-qwen3-vl-235b-a22b-instruct-fp8",
+  "base_url_configured": true,
+  "base_url_env": null,
+  "api_key_env": null,
+  "api_key_count": 0,
+  "network_calls_made": 0
+}

runs/20260615T211556Z_overnight-ready-smoke/logs/claim_ledger_runner_dry_run.json

@@ -0,0 +1,15 @@
+{
+  "schema": "meta3.runner_dry_run.v1",
+  "run_id": "20260615T211556Z_overnight-ready-smoke",
+  "mode": "claim_ledger",
+  "created_at": "2026-06-15T21:18:32.941272+00:00",
+  "execute_network": false,
+  "selected_packet_count": 1,
+  "concurrency": 2,
+  "model": "stackit-qwen-qwen3-vl-235b-a22b-instruct-fp8",
+  "base_url_configured": true,
+  "base_url_env": null,
+  "api_key_env": null,
+  "api_key_count": 0,
+  "network_calls_made": 0
+}

runs/20260615T211556Z_overnight-ready-smoke/manifests/baseline_snapshot.json

@@ -0,0 +1,43 @@
+{
+  "schema": "meta3.baseline_snapshot.v1",
+  "run_id": "20260615T211556Z_overnight-ready-smoke",
+  "created_at": "2026-06-15T21:16:04.298259+00:00",
+  "baseline_project": "/home/smlflg/Projekte/Meta2.0",
+  "source_artifacts": {
+    "meta2_root": "/home/smlflg/Projekte/Meta2.0",
+    "session_inventory": "/home/smlflg/Projekte/Meta2.0/data/session_inventory.json",
+    "layer1_digests": "/home/smlflg/Projekte/Meta2.0/data/layer1/session_digests.jsonl",
+    "layer1_audit": "/home/smlflg/Projekte/Meta2.0/data/layer1/audit.json"
+  },
+  "source_snapshot": {
+    "inventory_total": 1104,
+    "layer1_processed": 216,
+    "layer1_coverage_pct": 19.57,
+    "layer1_audit_digest_rows": 216,
+    "layer1_audit_unique_digest_paths": 216,
+    "layer1_raw_digest_file_rows_counted": 216,
+    "unresolved_failures": 0,
+    "raw_failure_rows": 6,
+    "by_source": {
+      "claude_projects": 109,
+      "codex_sessions": 103,
+      "hermes_tirith_logs": 3,
+      "hermes_codex_sessions": 1
+    },
+    "confidence": {
+      "high": 156,
+      "medium": 48,
+      "low": 12
+    },
+    "strategic_relevance": {
+      "infrastructure": 151,
+      "projects": 133,
+      "hai": 63,
+      "harness": 19,
+      "overload": 17,
+      "monetization": 11,
+      "security": 2
+    }
+  },
+  "network_calls_made": 0
+}

runs/20260615T211556Z_overnight-ready-smoke/manifests/batch_plan.json

@@ -0,0 +1,151 @@
+{
+  "schema": "meta3.batch_plan.v1",
+  "run_id": "20260615T211556Z_overnight-ready-smoke",
+  "created_at": "2026-06-15T21:16:04.298675+00:00",
+  "network_calls_made": 0,
+  "source_snapshot": {
+    "inventory_total": 1104,
+    "layer1_processed": 216,
+    "layer1_coverage_pct": 19.57,
+    "layer1_audit_digest_rows": 216,
+    "layer1_audit_unique_digest_paths": 216,
+    "layer1_raw_digest_file_rows_counted": 216,
+    "unresolved_failures": 0,
+    "raw_failure_rows": 6,
+    "by_source": {
+      "claude_projects": 109,
+      "codex_sessions": 103,
+      "hermes_tirith_logs": 3,
+      "hermes_codex_sessions": 1
+    },
+    "confidence": {
+      "high": 156,
+      "medium": 48,
+      "low": 12
+    },
+    "strategic_relevance": {
+      "infrastructure": 151,
+      "projects": 133,
+      "hai": 63,
+      "harness": 19,
+      "overload": 17,
+      "monetization": 11,
+      "security": 2
+    }
+  },
+  "baseline_snapshot_path": "/home/smlflg/Projekte/Meta3.0/runs/20260615T211556Z_overnight-ready-smoke/manifests/baseline_snapshot.json",
+  "settings": {
+    "safe_batch_size": 3,
+    "safe_concurrency": 2,
+    "question_count": 10,
+    "answer_passes_per_question": 3,
+    "synthesis_passes": 2,
+    "max_api_calls": 10000000
+  },
+  "process_model": {
+    "phase_count": 9,
+    "source": "/home/smlflg/Projekte/Meta3.0/config/process_model.json"
+  },
+  "estimated_calls": {
+    "layer1_backfill": 296,
+    "claim_ledger_generation": 30,
+    "owner_synthesis": 2,
+    "total": 328
+  },
+  "phases": [
+    {
+      "id": "p00_requirements_lock",
+      "kind": "manual_gate",
+      "status": "open",
+      "purpose": "Lock the owner question, allowed sources, stop conditions, budget, and morning success definition.",
+      "exit_gate": "Questions, constraints, and success criteria are reconciled.",
+      "blocking_reason": "requirements/PENDING.md has not been locked."
+    },
+    {
+      "id": "p01_baseline_audit",
+      "kind": "measurement_gate",
+      "status": "complete",
+      "purpose": "Measure the Meta2.0 baseline before changing scale.",
+      "exit_gate": "Coverage, unique digest count, raw row count, failures, and source spread are recorded.",
+      "inventory_total": 1104,
+      "coverage_pct": 19.57
+    },
+    {
+      "id": "p02_calibration_slice",
+      "kind": "small_model_run",
+      "status": "blocked_until_requirements_lock",
+      "purpose": "Answer a small representative slice before broad overnight execution.",
+      "exit_gate": "At least one answer shows evidence, uncertainty, counterevidence status, and owner-ready conclusion.",
+      "estimated_api_calls": 0
+    },
+    {
+      "id": "p03_stratified_backfill",
+      "kind": "meta2_layer1_backfill",
+      "status": "planned",
+      "purpose": "Increase digest coverage while preserving source and time diversity.",
+      "exit_gate": "New audit shows no unresolved failures and improved coverage.",
+      "remaining_sessions": 888,
+      "estimated_api_calls": 296,
+      "recommended_command": "cd /home/smlflg/Projekte/Meta2.0 && python3 scripts/meta2_layer1_digest.py --limit 120 --batch-size 3 --concurrency 2 --order source-balanced"
+    },
+    {
+      "id": "p04_claim_ledger_generation",
+      "kind": "structured_answer_run",
+      "status": "blocked_until_requirements_calibration_and_coverage",
+      "purpose": "Generate structured claims per question instead of freeform essays.",
+      "exit_gate": "Claims cite evidence and mark uncertainty.",
+      "estimated_api_calls": 30
+    },
+    {
+      "id": "p05_counterevidence_pass",
+      "kind": "adversarial_review",
+      "status": "blocked_until_claim_ledger",
+      "purpose": "Search for sessions that weaken, contradict, or narrow the generated claims.",
+      "exit_gate": "Each main claim has counterevidence status.",
+      "estimated_api_calls": 0
+    },
+    {
+      "id": "p06_validity_review",
+      "kind": "quality_gate",
+      "status": "blocked_until_claim_and_counterevidence_ledgers",
+      "purpose": "Separate useful answers from invalid measurement.",
+      "exit_gate": "Invalid, weak, and owner-ready outputs are explicitly separated.",
+      "estimated_api_calls": 0
+    },
+    {
+      "id": "p07_owner_synthesis",
+      "kind": "owner_report",
+      "status": "blocked_until_validity_review",
+      "purpose": "Compress validated claims into canonical owner-facing answers.",
+      "exit_gate": "Each question has one thesis, evidence, limits, and next action.",
+      "estimated_api_calls": 2
+    },
+    {
+      "id": "p08_morning_closeout",
+      "kind": "closure",
+      "status": "planned_after_run_end",
+      "purpose": "Make the overnight run inspectable and stop hidden cycle debt.",
+      "exit_gate": "Actual calls, errors, outputs, missing evidence, and next decision are visible.",
+      "estimated_api_calls": 0
+    }
+  ],
+  "gates": [
+    {
+      "gate": "layer1_coverage_below_final_answer_threshold",
+      "status": "open",
+      "coverage_pct": 19.57,
+      "required_pct": 90.0,
+      "meaning": "Final answers should wait; layer1 backfill can still run."
+    },
+    {
+      "gate": "dry_run_required_before_network_calls",
+      "status": "open",
+      "meaning": "This scaffold did not run a network dry-run."
+    },
+    {
+      "gate": "network_calls_disabled_by_default",
+      "status": "open",
+      "meaning": "A future runner must require an explicit start flag."
+    }
+  ]
+}

runs/20260615T211556Z_overnight-ready-smoke/manifests/run_manifest.json

@@ -0,0 +1,122 @@
+{
+  "schema": "meta3.run_manifest.v1",
+  "run_id": "20260615T211556Z_overnight-ready-smoke",
+  "created_at": "2026-06-15T21:15:56.851139+00:00",
+  "root": "/home/smlflg/Projekte/Meta3.0",
+  "mode": "prepare_only",
+  "profile_path": "/home/smlflg/Projekte/Meta3.0/config/run_profile.example.json",
+  "questions_path": "/home/smlflg/Projekte/Meta3.0/config/questions.json",
+  "process_model_path": "/home/smlflg/Projekte/Meta3.0/config/process_model.json",
+  "experiment_design_path": "/home/smlflg/Projekte/Meta3.0/config/experiment_design.json",
+  "question_count": 10,
+  "question_ids": [
+    "q01_roter_faden",
+    "q02_eigentliches_problem",
+    "q03_wiederkehrende_schleife",
+    "q04_staerkste_beweise",
+    "q05_selbsttaeuschungen",
+    "q06_beste_arbeitsmodi",
+    "q07_ownership_test",
+    "q08_produktlinie",
+    "q09_unfertige_wahrheiten",
+    "q10_finale_verdichtung"
+  ],
+  "process_phase_count": 9,
+  "process_phase_ids": [
+    "p00_requirements_lock",
+    "p01_baseline_audit",
+    "p02_calibration_slice",
+    "p03_stratified_backfill",
+    "p04_claim_ledger_generation",
+    "p05_counterevidence_pass",
+    "p06_validity_review",
+    "p07_owner_synthesis",
+    "p08_morning_closeout"
+  ],
+  "profile": {
+    "version": 1,
+    "name": "meta3_overnight_profile_v0",
+    "mode": "prepare_only",
+    "source_artifacts": {
+      "meta2_root": "/home/smlflg/Projekte/Meta2.0",
+      "session_inventory": "/home/smlflg/Projekte/Meta2.0/data/session_inventory.json",
+      "layer1_digests": "/home/smlflg/Projekte/Meta2.0/data/layer1/session_digests.jsonl",
+      "layer1_audit": "/home/smlflg/Projekte/Meta2.0/data/layer1/audit.json"
+    },
+    "api_profile": {
+      "provider": "openai_compatible_litellm",
+      "default_base_url": "https://litellm-kommone.genai.govdigital.de/v1",
+      "base_url_envs": [
+        "LITELLM_BASE_URL",
+        "OPENAI_BASE_URL"
+      ],
+      "api_key_envs": [
+        "LITELLM_API_KEYS",
+        "OPENAI_API_KEY"
+      ],
+      "model_env": "META3_MODEL",
+      "default_model": "stackit-qwen-qwen3-vl-235b-a22b-instruct-fp8"
+    },
+    "limits": {
+      "max_api_calls": 10000000,
+      "safe_batch_size": 3,
+      "safe_concurrency": 2,
+      "max_concurrency_without_new_approval": 4,
+      "dry_run_required": true,
+      "network_calls_enabled_by_default": false
+    },
+    "planning": {
+      "layer1_backfill_enabled": true,
+      "answer_passes_per_question": 3,
+      "synthesis_passes": 2,
+      "minimum_layer1_coverage_pct_before_final_answers": 90.0
+    }
+  },
+  "process_model_schema": "meta3.process_model.v1",
+  "experiment_design_schema": "meta3.experiment_design.v1",
+  "questions_version": 1,
+  "source_status": {
+    "meta2_root": {
+      "path": "/home/smlflg/Projekte/Meta2.0",
+      "exists": true,
+      "is_file": false,
+      "is_dir": true,
+      "size_bytes": null
+    },
+    "session_inventory": {
+      "path": "/home/smlflg/Projekte/Meta2.0/data/session_inventory.json",
+      "exists": true,
+      "is_file": true,
+      "is_dir": false,
+      "size_bytes": 615014
+    },
+    "layer1_digests": {
+      "path": "/home/smlflg/Projekte/Meta2.0/data/layer1/session_digests.jsonl",
+      "exists": true,
+      "is_file": true,
+      "is_dir": false,
+      "size_bytes": 298916
+    },
+    "layer1_audit": {
+      "path": "/home/smlflg/Projekte/Meta2.0/data/layer1/audit.json",
+      "exists": true,
+      "is_file": true,
+      "is_dir": false,
+      "size_bytes": 795
+    }
+  },
+  "env_status": {
+    "api_key_envs_present": {
+      "LITELLM_API_KEYS": false,
+      "OPENAI_API_KEY": false
+    },
+    "base_url_envs_present": {
+      "LITELLM_BASE_URL": false,
+      "OPENAI_BASE_URL": false
+    },
+    "model_env_present": false,
+    "secrets_recorded": false
+  },
+  "network_calls_made": 0,
+  "next_step": "Run scripts/meta3_plan_batches.py --run-id 20260615T211556Z_overnight-ready-smoke"
+}

runs/20260615T211556Z_overnight-ready-smoke/outputs/calibration_answers.jsonl

@@ -0,0 +1 @@
+{"answer": {"canonical_thesis": "Mock thesis for parser validation only.", "claims": [{"claim": "Mock claim anchored to the first digest.", "confidence": "low", "counterevidence_or_limit": "Mock response is not evidence for the real analysis.", "supporting_digest_indices": [1], "supporting_source_paths": ["/home/smlflg/.claude/projects/-home-smlflg--hermes-profiles-intake-router-hermes/52466b70-4d7e-4df7-9c62-bd20e18e478d.jsonl"]}], "counterevidence_status": "mock_not_evaluated", "mode": "calibration", "owner_next_step": "Run real network packets only after requirements and preflight pass.", "question_id": "q01_roter_faden", "uncertainty": "Synthetic response; validates extraction and schema only."}, "collected_at": "2026-06-15T21:26:09.646399+00:00", "mock": true, "mode": "calibration", "packet_id": "calibration__q01_roter_faden__shard-0001", "question_id": "q01_roter_faden", "question_title": "Der rote Faden", "response_path": "/home/smlflg/Projekte/Meta3.0/runs/20260615T211556Z_overnight-ready-smoke/outputs/responses/calibration/calibration__q01_roter_faden__shard-0001.json", "run_id": "20260615T211556Z_overnight-ready-smoke", "schema": "meta3.parsed_answer.v1"}

runs/20260615T211556Z_overnight-ready-smoke/outputs/parsed/calibration_parsed.jsonl

@@ -0,0 +1 @@
+{"answer": {"canonical_thesis": "Mock thesis for parser validation only.", "claims": [{"claim": "Mock claim anchored to the first digest.", "confidence": "low", "counterevidence_or_limit": "Mock response is not evidence for the real analysis.", "supporting_digest_indices": [1], "supporting_source_paths": ["/home/smlflg/.claude/projects/-home-smlflg--hermes-profiles-intake-router-hermes/52466b70-4d7e-4df7-9c62-bd20e18e478d.jsonl"]}], "counterevidence_status": "mock_not_evaluated", "mode": "calibration", "owner_next_step": "Run real network packets only after requirements and preflight pass.", "question_id": "q01_roter_faden", "uncertainty": "Synthetic response; validates extraction and schema only."}, "collected_at": "2026-06-15T21:26:09.646399+00:00", "mock": true, "mode": "calibration", "packet_id": "calibration__q01_roter_faden__shard-0001", "question_id": "q01_roter_faden", "question_title": "Der rote Faden", "response_path": "/home/smlflg/Projekte/Meta3.0/runs/20260615T211556Z_overnight-ready-smoke/outputs/responses/calibration/calibration__q01_roter_faden__shard-0001.json", "run_id": "20260615T211556Z_overnight-ready-smoke", "schema": "meta3.parsed_answer.v1"}

runs/20260615T211556Z_overnight-ready-smoke/outputs/responses/calibration/calibration__q01_roter_faden__shard-0001.json

@@ -0,0 +1,19 @@
+{
+  "packet_id": "calibration__q01_roter_faden__shard-0001",
+  "status_code": 200,
+  "started_at": "2026-06-15T21:26:03.261443+00:00",
+  "finished_at": "2026-06-15T21:26:03.261459+00:00",
+  "request_model": "mock-model",
+  "mock": true,
+  "network_calls_made": 0,
+  "response": {
+    "choices": [
+      {
+        "message": {
+          "role": "assistant",
+          "content": "{\"question_id\": \"q01_roter_faden\", \"mode\": \"calibration\", \"canonical_thesis\": \"Mock thesis for parser validation only.\", \"claims\": [{\"claim\": \"Mock claim anchored to the first digest.\", \"supporting_digest_indices\": [1], \"supporting_source_paths\": [\"/home/smlflg/.claude/projects/-home-smlflg--hermes-profiles-intake-router-hermes/52466b70-4d7e-4df7-9c62-bd20e18e478d.jsonl\"], \"counterevidence_or_limit\": \"Mock response is not evidence for the real analysis.\", \"confidence\": \"low\"}], \"counterevidence_status\": \"mock_not_evaluated\", \"uncertainty\": \"Synthetic response; validates extraction and schema only.\", \"owner_next_step\": \"Run real network packets only after requirements and preflight pass.\"}"
+        }
+      }
+    ]
+  }
+}

runs/20260615T211556Z_overnight-ready-smoke/outputs/workpackets/calibration_packet_manifest.json

@@ -0,0 +1,25 @@
+{
+  "schema": "meta3.workpacket_manifest.v1",
+  "run_id": "20260615T211556Z_overnight-ready-smoke",
+  "mode": "calibration",
+  "created_at": "2026-06-15T21:16:04.336420+00:00",
+  "network_calls_made": 0,
+  "source_digest_path": "/home/smlflg/Projekte/Meta2.0/data/layer1/session_digests.jsonl",
+  "available_digest_rows": 216,
+  "packet_count": 10,
+  "planned_call_count": 10,
+  "sample_size": 24,
+  "shard_size": 24,
+  "question_count": 10,
+  "unique_digest_refs_in_packets": 24,
+  "source_family_mentions": {
+    "claude_projects": 100,
+    "codex_sessions": 100,
+    "hermes_codex_sessions": 10,
+    "hermes_tirith_logs": 30
+  },
+  "packet_path": "/home/smlflg/Projekte/Meta3.0/runs/20260615T211556Z_overnight-ready-smoke/outputs/workpackets/calibration_workpackets.jsonl",
+  "call_ledger_path": "/home/smlflg/Projekte/Meta3.0/runs/20260615T211556Z_overnight-ready-smoke/state/calibration_call_ledger.jsonl",
+  "expected_response": "strict JSON per packet.expected_output_contract",
+  "next_step": "A future network runner may consume this packet file only after explicit approval."
+}

runs/20260615T211556Z_overnight-ready-smoke/outputs/workpackets/claim_ledger_packet_manifest.json

@@ -0,0 +1,26 @@
+{
+  "schema": "meta3.workpacket_manifest.v1",
+  "run_id": "20260615T211556Z_overnight-ready-smoke",
+  "mode": "claim_ledger",
+  "created_at": "2026-06-15T21:16:04.438524+00:00",
+  "network_calls_made": 0,
+  "source_digest_path": "/home/smlflg/Projekte/Meta2.0/data/layer1/session_digests.jsonl",
+  "available_digest_rows": 216,
+  "packet_count": 90,
+  "planned_call_count": 90,
+  "sample_size": 24,
+  "shard_size": 24,
+  "question_count": 10,
+  "unique_digest_refs_in_packets": 216,
+  "source_family_mentions": {
+    "claude_projects": 1090,
+    "codex_sessions": 1030,
+    "hermes_codex_sessions": 10,
+    "hermes_tirith_logs": 30
+  },
+  "packet_path": "/home/smlflg/Projekte/Meta3.0/runs/20260615T211556Z_overnight-ready-smoke/outputs/workpackets/claim_ledger_workpackets.jsonl",
+  "call_ledger_path": "/home/smlflg/Projekte/Meta3.0/runs/20260615T211556Z_overnight-ready-smoke/state/claim_ledger_call_ledger.jsonl",
+  "expected_response": "strict JSON per packet.expected_output_contract",
+  "next_step": "A future network runner may consume this packet file only after explicit approval.",
+  "shard_count_per_question": 9
+}

runs/20260615T211556Z_overnight-ready-smoke/reports/CALIBRATION_COLLECTION_REPORT.md

@@ -0,0 +1,15 @@
+# Meta3 Response Collection Report
+
+Run: `20260615T211556Z_overnight-ready-smoke`
+Mode: `calibration`
+Created: `2026-06-15T21:26:09.646775+00:00`
+Status: `pass`
+
+## Counts
+
+- Response files: 1
+- Parsed valid: 1
+- Invalid: 0
+- Missing responses: 9
+- Mock responses: 1
+- Network calls made by collector: 0