MICROSOFT AGENT FRAMEWORK · 0.16.0-rc.1
Demo030 — Agent Capabilities
Qualify five capabilities of the selected profile.
Objectives
- Qualify the actual provider capabilities before depending on them.
Prepare the demo
Start at the root of the complete solution already built in Release. Course prerequisites: Demo020. This program uses live chat from the selected profile; no embeddings are required except for Demo020 --embed. Check connectivity and required capabilities before attributing failure to the scenario.
| Project | Role |
|---|---|
| demos/Demo030_AgentCapabilities/Demo030.App.csproj | App |
Launch from the root
dotnet run --project demos/Demo030_AgentCapabilities/Demo030.App.csproj -c Release --no-builddotnet run --project demos/Demo030_AgentCapabilities/Demo030.App.csproj -c Release --no-build -- --Report capability-report.jsonModes and procedure
This program sends live text, structure, tool, streaming and cancellation requests. It needs no interactive input. Allow for its ten-minute overall deadline and two-minute probe deadlines; Ctrl+C requests cancellation.
var report = new
{
checkedAt = DateTimeOffset.UtcNow,
provider = settings.ChatProvider.ToString(),
model = profile.Model,
reasoningEffort = configuration["Pilot:ReasoningEffort"] ?? "provider default",
protocol = settings.ChatProvider == ConnectionProvider.Ollama ? "Ollama /api/chat" : "OpenAI /v1/chat/completions",
results
};
var json = JsonSerializer.Serialize(report, new JsonSerializerOptions { WriteIndented = true });
Console.WriteLine(json);
Results and useful failures
Read report JSON: checkedAt, provider, model, protocol, reasoningEffort and five results. Exit status is zero only when all five probes pass. The tool must actually produce an unpredictable receipt that reaches the response. Usage may stay null. --Report writes the report to the supplied path; --Pilot:ReasoningEffort none explicitly changes the reasoning setting.
These observations are expected from inspecting the indicated code revision. No new model call was made to write this page; exact generated wording is not a reference output.
Scope and limits
A Failed probe is an observed limitation, not partial success to hide. Protocol support and passing probes qualify neither every model nor a business judge.
