Tests Overview

Why you're reading this page: This page gives a short overview of Intentum test projects (Intentum.Tests, Intentum.Tests.Integration): what is covered, how to run them, and how they relate to samples and examples. It is the right place if you are asking "What does each test do?"

This page gives a short overview of Intentum tests: what is covered, how to run them, and how they relate to samples and examples.

Test projects

Project	Description
Intentum.Tests	Unit and contract tests: BehaviorSpace, inference, policy, providers (mock HTTP), clustering, explainability, simulation, versioning, multi-tenancy, events, experiments. No real API keys.
Intentum.Tests.Integration	Integration tests: greenwashing case study (accuracy/F1 on labeled data). Optional: real APIs if env vars set.

How to run

From the repository root:

# All unit tests (no API keys)
dotnet test tests/Intentum.Tests/Intentum.Tests.csproj

# Exclude provider integration tests (OpenAI, Azure, Gemini) when no keys
dotnet test tests/Intentum.Tests/Intentum.Tests.csproj --filter "FullyQualifiedName!=Intentum.Tests.OpenAIIntegrationTests&FullyQualifiedName!=Intentum.Tests.AzureOpenAIIntegrationTests&FullyQualifiedName!=Intentum.Tests.GeminiIntegrationTests&FullyQualifiedName!=Intentum.Tests.MistralIntegrationTests"

# Integration tests
dotnet test tests/Intentum.Tests.Integration/Intentum.Tests.Integration.csproj

See Testing for details and Local integration tests for scripts.

What is covered (summary)

Core: BehaviorSpace, ToVector, intent confidence, policy engine (Evaluate, EvaluateWithRule), rate limit, localization; intent resolution pipeline; behavior space sanitization (anonymization/masking).
Models: Rule-based, chained, multi-stage, sliding window, LLM (mock + provider parsing with mock HTTP); ONNX intent model (constructor/options).
Analytics: IntentAnalytics (trends, distribution, anomalies, timeline, export, intent graph snapshot).
Explainability: IntentExplainer, IntentTreeExplainer, decision tree.
Observability: Policy execution log (DecideWithExecutionLog), metrics.
Persistence: In-memory repo (history + behavior space), EF/Redis/Mongo not required for unit tests.
Simulation: BehaviorSpaceSimulator, ScenarioRunner.
Streaming: BoundedMemoryBehaviorStreamConsumer, WindowedBatchBuffer.
Patterns: BehaviorPatternDetector, template matching.
Policy store: FilePolicyStore, SafeConditionBuilder (declarative rules).

Integration tests cover greenwashing accuracy/F1 on labeled data; see Case study — Greenwashing metrics.

Tests vs samples vs examples

	Tests	Examples	Samples
Purpose	Assert contracts and core behavior; no real APIs in CI.	Learn one use case; copy-paste friendly.	Full app: many features, Web API, UI.
Run	`dotnet test tests/Intentum.Tests`	`dotnet run --project examples/<name>`	`dotnet run --project samples/Intentum.Sample.Blazor`
Docs	Testing, this page	Examples overview	API, Setup

Adding tests

Use mock HttpClient or in-memory providers so tests don’t call real APIs.
For new features (e.g. timeline, intent tree, pattern detector), add tests in Intentum.Tests and extend the “What is covered” list in Testing.