feat/phase-2g-bundle
Phase 2: Indexer — chunk, embed, vector store (Qdrant)
- lib/chunk.mjs: Markdown chunking by heading then size/overlap; stable chunk id - lib/embedding.mjs: Ollama (nomic-embed-text) and OpenAI; embeddingDimension() - lib/vector-store.mjs: Qdrant backend; ensureCollection, upsert (idempotent) - lib/indexer.mjs: runIndex() — vault list, chunk, embed batches, upsert - CLI knowtation index and scripts/index-vault.mjs both run indexer - Config: embedding + indexer chunk_size/chunk_overlap defaults - IMPLEMENTATION-PLAN.md: Phase 2 marked complete; sqlite-vec noted as future
Requires: config/local.yaml or env KNOWTATION_VAULT_PATH, QDRANT_URL. Ollama required for default embedding (or set embedding.provider/model).
sha256:9744eb0d1897d088a7ca616b528edbc2756a5f4bf0cbe84ae5e7d757b8f56d85
sha
sha256:d4e0278381a8c0eb62f02e37fd3ae98d24a17cd6439aa8845f1c25e5ed4267c9
snapshot
0 comments
To add a comment, use the Muse CLI:
muse hub commit comment sha256:9744eb0d1897d088a7ca616b528edbc2756a5f4bf0cbe84ae5e7d757b8f56d85 --body "your comment"
No comments yet. Be the first to start the discussion.