DEFAULT_EXTRACTION_CONCURRENCY

@forge/monorepo / backend/src / DEFAULT_EXTRACTION_CONCURRENCY

Variable: DEFAULT_EXTRACTION_CONCURRENCY

const DEFAULT_EXTRACTION_CONCURRENCY: 1 = 1

Defined in: backend/src/knowledge/graph.ts:338

Extraction runs one chunk at a time by default.

Sequential rather than parallel because the cost is the point: a corpus of 603 chunks is 603 model calls, and firing them concurrently is how a re-index becomes a rate-limit incident. A deployment that wants throughput raises this deliberately, having read the number.