Edgedb

EdgeDB is a next-generation relational database with a graph-relational data model and strongly typed schema that aims to replace ORMs by making the database itself expressive for complex relational queries. EdgeQL (EdgeDB's query language) can express in a single query what would require multiple ORM operations or complex SQL JOINs. The HTTP API and GraphQL API enable agents to issue queries without persistent database connections — well-suited for serverless and stateless agent contexts that can't maintain connection pools.

Authentication uses secret keys for the HTTP API and EdgeDB's built-in access control for client connections. The managed cloud uses token-based auth for the HTTP API. EdgeDB's access policies (defined in schema) enable row-level security enforced at the database layer.

Documentation is excellent and developer-focused. The EdgeQL reference is thorough and the schema definition documentation explains the type system well. Teams evaluating EdgeDB as a database choice should invest in the EdgeQL learning curve — it is different from SQL but expressive for complex object-relational queries that would be painful in standard SQL.

The API covers EdgeQL queries via HTTP POST and a GraphQL endpoint auto-generated from the schema. The HTTP API accepts EdgeQL expressions and returns JSON — agents can run complex relational queries including computed properties, link traversal, and nested object construction without assembling JOIN results. EdgeDB Cloud provides the managed hosting option with the same HTTP API surface.

Reliability for EdgeDB Cloud is vendor-managed. Self-hosted EdgeDB reliability depends on the deployment configuration and the underlying Postgres storage layer. EdgeDB's Postgres-backed storage inherits the reliability characteristics of a battle-tested database engine while adding the EdgeDB query layer.