- Neue Tastaturzeile: Unterlagen AUS/AN zeigt Modus und schaltet um.
- document_mode in ask_with_tools: erzwingt lokales Modell und RAG-Pflicht
wie bei Doc-Keywords (Session wird bei Suche wie bisher bereinigt).
- Optional: doku:/rag: Prefix fuer einmalige Suche ohne Modus.
- Sprache und Hilfetext ergaenzt.
- Remove standalone "kosten" from WEB_TRIGGERS (matched "jährlichen kosten" and sent Perplexity without tools).
- Add LOCAL_OVERRIDES for jahreskosten/jaehrlich/beitragsrechnung phrases.
- Extend forced-RAG keywords and enrich ES query with Beitrag/Beitragsrechnung for cost questions.
- System hint: extract amounts and periods per hit when possible.
- Forced rag_search used top_k=10 and tool content up to 12k chars (was 3k, cut off most hits).
- System instruction: enumerate each Sparte/doc type for same insurer, not only first hit.
- rag.py SYSTEM_PROMPT_EXTRA: same rule for multi-hit answers.
Poisoned session history (4x wrong answers) overrode RAG results.
Now: when doc keywords trigger forced rag_search, rebuild messages
from scratch with ONLY system prompt + RAG results + question.
No session history = no poisoning.
The LLM ignores system injections when session history already
contains insurance answers. Now: when doc keywords detected, call
rag_search directly in Python, inject results as fake tool-call
into messages. LLM only needs to format the answer.
When session history already contained insurance/document answers,
the LLM would skip rag_search and hallucinate from memory.
Now: if doc keywords detected, inject system-message reminder AFTER
session history but BEFORE user question, forcing rag_search call.
Fixes#51 follow-up: LVM query returned wrong answer.
Versicherung/Vertrag/Steuer/Rente/Dokument etc. routen jetzt lokal
statt zu Sonar, damit rag_search per Tool-Calling aufgerufen wird.
Vorher: aktuell in Frage -> Sonar -> kein Tool-Calling -> kein RAG.
Ref: Issue #51
