What is a Source Adapter?
A source adapter is the section of a capability spec that declares how the capability connects to one or more upstream APIs. It defines the base URI, authentication method, available operations, and the abstraction level used for each consumed system.
Types of source adapters
- Pass-through (L1) – Forwards requests directly to the upstream API with minimal transformation.
- Templatized (L2) – Uses URI templates and variable injection to parameterize upstream calls.
- Structured (L3) – Full request/response modeling with typed parameters and JSONPath mapping.
- Functional (L4) – AI-native interface with semantic descriptions optimized for MCP tool discovery.
Reusability
Source adapter declarations can be shared across capabilities using external references, so teams define a connection to an upstream API once and reuse it across multiple capabilities. This prevents configuration drift and reduces the overhead of managing upstream API changes.
Why it matters
Source adapters abstract the complexity of upstream API consumption into a declarative contract. Teams describe what they need from an API, and the framework handles connection management, authentication, retries, and error handling.
