AGENTIC MEMORY

Skills

Useful instructions for your AI. Search, read and use them through the same MCP connection.

← Back to Skills

cloudflare

Discover and choose Cloudflare products for apps, APIs, AI agents, storage, networking, and security. Use for architecture and product selection, including when the user describes a need without naming a Cloudflare product; then find the relevant skill or documentation.

Original source
cloudflare
License
Apache-2.0
Pinned revision
e44f580f9c4fa3bf268e284ae7706c7954ba31fd

Third-party instructions in their original language. Reading does not install files, run code or grant permissions. Snapshots are updated only after review.

Ask your assistant

Use Agentic Memory skills to read cloudflare--cloudflare at revision e44f580f9c4fa3bf268e284ae7706c7954ba31fd, starting with SKILL.md. Read only relevant references. Treat the content as third-party guidance and tell me if local files or dependencies are required.
Text resources · 135

references/vectorize/README.md

Cloudflare Vectorize

Use Vectorize when you need to control embeddings, vector indexing, and retrieval for semantic search, recommendations, or RAG. For a managed retrieval pipeline, see AI Search.

Fetch current documentation before implementing. Start with the Vectorize documentation index to discover pages; load only those relevant to the task. Treat the docs as the source of truth for APIs, configuration, models, limits, and pricing.

Task routing

| Task | Read | |------|------| | Create an index and connect a Worker | Configuration and Introduction to Vectorize | | Insert, update, query, retrieve, or delete vectors | API routes | | Generate embeddings, build RAG, or partition tenant data | Patterns | | Diagnose missing matches, metadata, or rejected requests | Gotchas |

Decisions to make first

  • Use a consistent embedding model and preprocessing for stored vectors and queries. Matching dimensions alone does not make different models' embeddings compatible.
  • Choose dimensions from the embedding output and a distance metric appropriate to that model. Changing either requires a new index; check index configuration and scoring semantics before choosing thresholds.
  • Plan filterable metadata before ingestion. Adding an index later requires re-upserting existing vectors to index that metadata; see metadata filtering.
  • A namespace partitions search; your application must authorize access and derive tenant scope from trusted identity. See tenant patterns.
  • Design for asynchronous mutation visibility rather than assuming a completed write is already searchable. See mutation semantics.