feat(knowledge): LLM 提炼陈述式知识(wiki-capture 规范),规则抽取兜底

之前沉淀直接堆砌对话原文。现默认用大模型按 wiki-capture 规范提炼为 Context/Finding/Reasoning/Implications/Related 结构化陈述式知识,过滤寒暄与中间推理;LLM 不可用或解析失败时回退规则抽取。新增 llm_extractor.py + 配置 llm_extract_enabled/extract_model_name/extract_max_input_chars。

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This commit is contained in:
AdministratorandClaude Opus 4.8 committed 2026-06-08 18:00:46 +08:00
1 parent d06c8d49d7
commit 8463289f75
4 files changed
+246 -12

No files matched your search

@@ -46,6 +46,9 @@ class KnowledgeConfig(BaseModel):
project_name: str = Field(default="deerflow", description="Project namespace for projects/<name>/ pages.")
auto_ingest_enabled: bool = Field(default=False, description="Auto-sediment conversations in the background (phase 3).")
auto_ingest_min_chars: int = Field(default=300, description="Minimum characters before auto-ingest considers a turn.")
llm_extract_enabled: bool = Field(default=True, description="Distill notes with an LLM (wiki-capture style) instead of the rule-based fallback.")
extract_model_name: str | None = Field(default=None, description="Model (from models[]) used to distill knowledge. Falls back to models[0] when unset.")
extract_max_input_chars: int = Field(default=12000, description="Max characters of conversation transcript fed to the extraction LLM.")
max_thread_messages: int = Field(default=80, description="Maximum thread messages to scan when capturing.")
max_source_items: int = Field(default=20, description="Maximum source items recorded per note.")
rag_enabled: bool = Field(default=False, description="Enable RAG injection before agent runs (phase 2).")