Docs & Quickstart
BaseKV supports Redis and Memcached client protocols, a DynamoDB-style API for simple item get/put workflows, and a RedisGraph-style command subset for property graph use cases. Start by creating a database, then point your client at the BaseKV endpoint. You can run the same workflow from a human dashboard today and move toward machine-managed automation.
Quickstart
- Create a database in the dashboard.
- Choose your access path: Redis, Memcached, DynamoDB-style, or Redis graph commands.
- Copy the endpoint details for that protocol.
- Update your client connection settings.
- Write a test key to confirm connectivity.
Graph Compatibility Subset
BaseKV exposes RedisGraph-style commands for teams that model data as nodes and edges. The mental model is close to Neo4j (property graph + Cypher-like queries), but the implementation is a compatibility subset and not full Neo4j or full RedisGraph parity.
Supported commands:
GRAPH.QUERYGRAPH.RO_QUERYGRAPH.EXPLAINGRAPH.PROFILEGRAPH.CONFIGGRAPH.LISTGRAPH.DELETE- Cypher subset includes CREATE, MERGE, MATCH, WHERE, SET, DELETE, RETURN, LIMIT, and basic relationship patterns.
- Read-only flows can use
GRAPH.RO_QUERY, which rejects write statements. - Graph keys integrate with Redis keyspace operations like
TYPE,EXISTS,DEL,KEYS, andSCAN.
Neo4j performance positioning (working estimate)
For deep traversals (3+ hops), current BaseKV graph execution is typically slower than Neo4j because traversals resolve adjacency through BoltDB index lookups rather than index-free pointer jumps.
- Small graphs (<10k nodes): often negligible user-visible differences.
- Medium graphs (10k-1M nodes): commonly around 5x-15x slower for traversal-heavy patterns.
- Deep traversal workloads: practical range is often 10x-50x slower today; at very large graph sizes this can exceed 50x.
Treat these numbers as planning guidance, not formal benchmarks. Always benchmark your own query mix before production commitments.
For Agents
If you are evaluating BaseKV for autonomous workflows, start with the agent quickstart and security model before scaling automation. Use the lifecycle API guide for task-by-task automation and the MCP status page for integration direction.
Machine-readable discovery endpoints are available for integration bootstrap and tooling.
Switching notes
For Redis users, swap your connection string and keep your client library. For Memcached, you can often keep the same client with a new endpoint. For DynamoDB-style integrations, BaseKV targets simple item get/put workflows rather than full DynamoDB feature parity. For graph workloads, BaseKV offers RedisGraph-style command compatibility with a practical subset and should be validated query-by-query. BaseKV focuses on compatibility you can validate, not opaque migrations.
See the compatibility page for a practical overview of what each integration is intended for.