agrag.ingestion.reports.add_result.AddResult

Bases: BaseModel

Return type of Graph.add(). One summary per pipeline stage.

Attributes:

• ingestion (IngestStats) – Ingestion-stage results.
• chunking (ChunkingStats) – The chunker each document got and the chunks it made.
• extraction (ExtractionStats) – Extractor output across every chunk this call processed.
• escalation (EscalationReport) – How many chunks the LLM extractor read, how many the budget left to the first extractor, how many escalations failed, and the wall time of each phase. Only the chunk count and the first-phase time are set when the graph has no LLM extractor.
• resolution (ResolutionStats) – Resolution's tier-by-tier match counts.
• merge (MergeStats) – What merge mechanics did with resolution's groups.
• storage (StorageStats) – What made it to GraphStore, and what didn't.
• chunks (list[Chunk]) – Every Chunk this call produced. Empty unless return_chunks=True — holding full chunk text for a large corpus is a real memory cost most callers don't need paid for.

chunking

chunking: ChunkingStats = Field(default_factory=ChunkingStats)

chunks

chunks: list[Chunk] = Field(default_factory=list)

escalation

escalation: EscalationReport = Field(default_factory=EscalationReport)

extraction

extraction: ExtractionStats = Field(default_factory=ExtractionStats)

ingestion

ingestion: IngestStats = Field(default_factory=IngestStats)

merge

merge: MergeStats = Field(default_factory=MergeStats)

resolution

resolution: ResolutionStats = Field(default_factory=ResolutionStats)

storage

storage: StorageStats = Field(default_factory=StorageStats)