Where the runnable code lives in this repository.
Conductor.AI.Examples/ — 176 self-contained example projects, numbered roughly by
topic and increasing in complexity:
| Range | Topic |
|---|---|
01_* |
Basic agent |
02_* |
Tools — simple, multi-step, registries |
03_* |
Structured output |
04_* |
HTTP and MCP tools |
Sk*_* |
Semantic Kernel adapter |
Each directory is its own Program.cs with a header comment listing the environment
variables it needs. Run one directly:
dotnet run --project Conductor.AI.Examples/01_BasicAgentThe CI job Build agent examples in pull_request.yml compiles all of them, so they
stay buildable.
csharp-examples/ — workflow and worker examples:
| File | Covers |
|---|---|
WorkFlowExamples.cs |
Workflow authoring and execution. |
HumanTaskExamples.cs |
Human task flows. |
TestWorker.cs |
A minimal worker. |
Runner.cs / Program.cs |
Entry points. |
There is a Dockerfile (and DockerfileMacArm) for running these in a container.
The test suites are often the most complete and most current reference, because CI keeps them honest:
| Location | Covers |
|---|---|
Tests/Worker/ |
Worker behaviour. |
Tests/Integration/ |
End-to-end workflow runs against a live server. |
Tests/Definition/ |
Workflow definition building. |
Tests/Telemetry/ |
Metrics collection. |
Conductor.AI.Tests/ |
Agent unit tests. |
Conductor.AI.E2eTests/ |
Numbered agent E2E suites — tool calling, CLI tools, HTTP tools, stateful domains, media, PLAN_EXECUTE, skills, agent client, auth headers, schedules. |
Conductor.AI.E2eTests/Suite17_SdkParity.cs is worth knowing about specifically: it
checks behaviour parity with the sibling SDKs.
They need a live server and provider credentials, and are driven by the Agent E2E
workflow (.github/workflows/agent-e2e.yml) rather than the ordinary CI build. That
workflow downloads a server jar, starts it, waits for health, and runs the suites — read
it for the exact environment expected.
Additional examples outside this repo: conductor-sdk/conductor-examples.