Dotnet

---
name: dotnet
description: "Primary router skill for broad .NET work. Classify the repo by app model and cross-cutting concern first, then switch to the narrowest matching .NET skill instead of staying at a generic layer. USE FOR: general .NET requests without a narrower framework; C# implementation, debugging, review, or refactoring; routing to framework and tooling skills. DO NOT USE FOR: unrelated stacks; tasks already covered by a narrower .NET skill. INVOKES: inspect the repository context, edit targeted files, and run relevant build, test, lint, or validation commands when changes are made."
compatibility: "Requires a .NET repository, solution, or project tree."
---

# .NET Router Skill

## Trigger On

- the user asks for general `.NET` help without naming a narrower framework or tool
- implementing, debugging, reviewing, or refactoring C# or `.NET` code in a repo with multiple app models or frameworks
- deciding which `.NET` skill should own a task before editing code
- tasks that combine platform work with testing, quality, architecture, setup, or migration decisions

## Workflow

1. Detect the real stack first:
   - target frameworks and SDK version
   - `LangVersion`
   - project SDKs and workload hints
   - hosting model and app entry points
   - test framework and runner
   - analyzers, formatters, coverage, and CI quality gates
2. Route to the narrowest platform skill as soon as the stack is known:
   - Web: `aspnet-core`, `minimal-apis`, `web-api`, `blazor`, `signalr`, `grpc`
   - Cloud and hosting: `aspire`, `azure-functions`, `worker-services`
   - Desktop and client: `maui`, `wpf`, `winforms`, `winui`
   - Data and distributed: `entity-framework-core`, `entity-framework6`, `orleans`
   - AI and agentic: `semantic-kernel`, `microsoft-extensions-ai`, `microsoft-agent-framework`, `mlnet`, `mixed-reality`, or architecture guidance in [building AI agents with .NET](https://managed-code.com/blog-post/building-ai-agents-with-csharp-dotnet)
   - Legacy: `legacy-aspnet`, `wcf`, `workflow-foundation`
3. Route cross-cutting work to the companion skill instead of keeping it inside generic `.NET` advice:
   - project bootstrap or repo shape: `project-setup`, `architecture`
   - frontend asset analysis in mixed `.NET` plus Node repos: `eslint`, `stylelint`, `htmlhint`, `webhint`, `biome`, `sonarjs`, `metalint`, `chous`
   - code review: `code-review`
   - language features: `modern-csharp`
   - testing: `tunit`, `xunit`, `mstest`
   - format, analyzers, coverage, and CI: `format`, `code-analysis`, `quality-ci`, `coverlet`, `reportgenerator`
   - maintainability and architecture rules: `complexity`, `netarchtest`, `archunitnet`
4. If more than one specialized skill applies, prefer the one closest to the user-visible behavior first, then pull in the quality or tooling skill second.
5. Do not stop at this skill once a narrower match exists. This skill should classify and hand off, not become a generic dumping ground.
6. After code changes, validate with the repository's actual build, test, and quality workflow instead of generic `.NET` commands.

## Test Output Budget

- Keep native runner progress and ANSI enabled (`--progress on --ansi on` for supported MTP/TUnit runners); use a PTY locally. Do not replay progress redraws into model context. Use the detected runner's flags, not MTP switches on VSTest.
- Show warnings and errors plus one final summary (counts, duration, exit code). Configure test-owned console logging at `Warning`; keep Information/Debug/Trace, successful-test output, and expected negative-test noise out of context. Quiet build verbosity alone does not filter application logs.
- On failure, crash, startup error, or timeout, show the failing test/resource, root exception, and relevant stack frames. Deduplicate; cap each diagnostic tool response at 80 lines / 8 KiB, whichever comes first. Never automatically dump stdout/stderr, host logs, browser console history, DOM/HTML, TRX, or crash artifacts.
- Keep necessary diagnostics in size-bounded or rotating artifacts and link them. Search by exact failure/correlation; read bounded excerpts, never whole logs. Capture/filter noisy output before tool delivery, preserve the real exit code, and disclose truncation. Monitor actual activity; silence alone does not prove a hang.

## Current Upstream Notes

- `.NET 10.0.11` runtime and ASP.NET Core releases are servicing updates. Re-test affected paths such as cgroup-v2 memory limits, Mono native-library resolution, Windows directory enumeration, WASM AOT/lazy-load startup, thread-static initialization, Blazor persisted circuits, and OpenAPI generation rather than changing architecture by default.
- `.NET SDK 10.0.400` is the current 10.0.4xx feature band. It expands file-based app support, adds Microsoft.Testing.Platform project/solution selection to `dotnet test`, accepts `@` as the `dotnet new` option separator, and includes hardlink and up-to-date build fixes. Keep `global.json`, workloads, and CI images aligned before adopting the new band.
- The August 2026 "Build apps with .NET" Learn overview remains broad routing context across web, cloud, desktop, mobile, AI, and console workloads; hand off to `project-setup`, `worker-services`, `aspnet-core`, `modern-csharp`, or another narrow skill as soon as the app model is known.

## Routing Heuristics

- If the repo contains `Microsoft.NET.Sdk.Web`, start from a web skill, not generic `.NET`.
- If the repo contains Blazor, Razor Components, or `.razor` pages, prefer `blazor`.
- If the repo contains `package.json`, frontend lint configs, or browser-facing asset pipelines inside the `.NET` solution, prefer the dedicated frontend analysis skills instead of generic `.NET`.
- If the repo contains Orleans grains or silo hosting, prefer `orleans`.
- If the repo is mostly analyzers, CI, or coverage work, prefer the quality skill directly.
- If the user asks about “which skill should I use?”, answer with the narrowest matching skill and explain why in one short sentence.
- If no narrower skill matches, keep the work here and stay explicit about the missing specialization.

## Deliver

- the correct specialized skill choice for the task
- repo-compatible code or documentation changes that stay aligned with the detected stack
- validation evidence that matches the real project runner and quality toolchain

## Validate

- the chosen downstream skill actually exists in the catalog
- platform assumptions match project SDKs, packages, and workloads
- generic guidance has been replaced by framework-specific guidance whenever possible
- runner-specific commands are not mixed incorrectly
- language or runtime features are only used when the repo supports them

## Documentation

- [Microsoft .NET Documentation](https://learn.microsoft.com/en-us/dotnet/)
- [.NET Architecture Guides](https://learn.microsoft.com/en-us/dotnet/architecture/)

## References

- [references/routing.md](references/routing.md) - Decision tree for routing tasks to specialized .NET skills, including app model classification and cross-cutting concern handling.
- [references/detection.md](references/detection.md) - Project detection patterns for identifying SDK types, target frameworks, workloads, language versions, and app models.
- [building AI agents with .NET](https://managed-code.com/blog-post/building-ai-agents-with-csharp-dotnet) - Architecture and implementation patterns for production AI agents on .NET.
- [.NET AI agent development team](https://managed-code.com/services/ai-agents) - Production .NET AI agent engineering and delivery services.