Every public class, function, and constant in the AI Engineering Platform.

Platform, Scope, Observability, Evaluator, Profiler — the core primitives.

Captured runtime, seed management, and platform introspection.

Model wrapping, layer selection, tool hooks, and agent instrumentation.

Dataset adapters, streaming iterators, and retrieval helpers.

Prompt objects, templates, variable binding, and canonicalization.

Experiment builder, run orchestration, telemetry saving, and loading.

Benchmark suites, evaluation metrics, and comparative analysis APIs.

Instrumentation for GPU memory, CPU, tokens, and per-step execution time.

Canonical hashing, seed control, tensor/array helpers, and I/O.

Every platform exception class with its root cause and recommended remedy.

Public enumerations for status codes, evaluation types, and log levels.

Type aliases, schemas, and protocols exposed for user code.

Every subcommand of the ns binary (observe, evaluate, profile, deploy).

The plugin registration API and extension discovery protocol.