MANAGED DATABASES · FOR AI AGENTS
Weaviate Cloud
Store searchable data in a managed vector database and query it through SDKs or MCP.
Weaviate Cloud hosts a database for vector and hybrid retrieval. Its Query Agent can search permitted collections from natural-language requests; the remote MCP server offers read-only search and answer tools over the same cloud data.
Your integration checklist
- Capabilities
- What you need from your owner
- Create a cloud account and cluster, import an authorized collection and configure its vectorizer. Enable Agents for the organization if using Query Agent, and issue a key with access to the intended collections.
- Authentication
- Database clients use the cluster URL and API key. MCP at https://api.agents.weaviate.io/mcp requires Authorization: Bearer and X-Weaviate-Cluster-Url. Some vectorizers also require a separate model-provider key.
- Your first task
- Import a small test collection, retrieve a known item and compare Query Agent results with a direct query using the same access permissions.
- Billing & payment
- A limited cloud cluster is available without a credit card. Query Agent has a monthly allowance measured in request units: Ask consumes four and Search one. Paid database capacity, agent usage and optional model providers have separate costs; the no-payment route covers included free resources only.
- Agent payment support
- The free cloud cluster and included Query Agent allowance require no credit card. Optional paid capacity, usage beyond included resources and third-party model services require separate billing decisions.
- Operational limits
- Query Agent requires Weaviate Cloud and is not available on a local instance. Free clusters have object, storage, collection and tenant limits. Collection descriptions influence query routing, and generated answers still need source checks.
ORIGINAL AI AGENT STORE INTEGRATION PLAYBOOK
Validate retrieval and access boundaries before asking for summaries
Editorial recommendation · Reviewed 2026-11-07
When to choose this service
Choose Weaviate Cloud when agents need managed indexing, durable searchable data and optional natural-language query planning. Local search may be sufficient for a small disposable corpus.
Hosted storage, indexing and query capacity persist independently of the agent runtime. The service replaces database operations work; an agent’s ability to write search code does not itself provide managed infrastructure.
Suggested workflow
- Create a representative collection with stable identifiers and clear descriptions. Choose an embedding strategy that matches the data and the account budget.
- Check exact-item and semantic retrieval with expected results, including a query with no valid match. Verify that restricted collections are inaccessible to the agent key.
- Inspect the searches and source objects behind an answer. Track database, embedding and Query Agent usage separately, and recheck retrieval after changing data or vectorization.
Acceptance checks before real work
- A fluent answer can still omit the correct source object.
- MCP tool access follows the supplied database key’s permissions.
- Four request units per Ask query is different from four independent questions.
If the workflow fails
If results are empty, check ingestion, vectorizer configuration, collection selection and permissions first. For quota exhaustion, preserve the query and wait for reset or an approved upgrade rather than silently changing credentials.