Embeddings (Vector Stores & Files)

Purpose: Build private knowledge bases and attach them to agents for retrieval‑augmented generation (RAG).

Typical flow

1. Upload files → PDF, Markdown, text, code, etc.
2. Create a Vector Store → choose a name and optionally a chunking strategy and embedding model.
3. Add files to the store → embedding runs asynchronously; monitor counts.
4. Attach the collection to your agent via Use Embeddings (RAG) in the Agents tab.
5. Test in Prompt Studio and verify that the agent cites injected snippets.

API parity

Files:

• GET /v1/files
• POST /v1/files (multipart)
• GET /v1/files/:id
• DELETE /v1/files/:id
• GET /v1/files/:id/content
  See Files

Vector Stores:

• GET /v1/vector_stores
• POST /v1/vector_stores
• GET /v1/vector_stores/:id
• DELETE /v1/vector_stores/:id
• GET /v1/vector_stores/:id/files
• POST /v1/vector_stores/:id/files
• GET /v1/vector_stores/:id/files/:fileId
• POST /v1/vector_stores/cancel
  See Vector Stores

Convenience:

• GET /v1/user-collections
• GET /v1/vector_stores/:id/file-details
  See Convenience

Diagnostics & Recovery (Mission Squad Core collections):

• GET /v1/core/collections
• POST /v1/core/collections/:collectionName/search
• GET /v1/core/collections/:collectionName/diagnostics
• POST /v1/core/collections/:collectionName/recover
  See Collections

Tips

• Keep files small and focused when possible for higher‑quality retrieval.
• Use consistent chunking strategy across similar document types.
• Re‑run diagnostics if retrieval quality drops after large data changes.