/ Directory / Playground / .NET Skills
● Community Aaronontheweb ⚡ Instant

.NET Skills

by Aaronontheweb · Aaronontheweb/dotnet-skills

Idiomatic .NET 8/9 in Claude — skills for xUnit testing, BenchmarkDotNet perf, NuGet packaging, async/await correctness, and C# style.

Aaron Stannard's .NET Skills are a battle-tested bundle for serious .NET developers. Beyond simple code generation, they cover the parts the model habitually gets wrong: async cancellation, ConfigureAwait usage, IDisposable, struct vs class tradeoffs, BenchmarkDotNet patterns, NuGet packaging metadata, and modern xUnit + FluentAssertions style.

Why use it

Key features

Live Demo

What it looks like in practice

ready

Install

Pick your client

~/Library/Application Support/Claude/claude_desktop_config.json  · Windows: %APPDATA%\Claude\claude_desktop_config.json
{
  "mcpServers": {
    "dotnet-skills-aaronontheweb": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/Aaronontheweb/dotnet-skills",
        "~/.claude/skills/dotnet"
      ],
      "_inferred": true
    }
  }
}

Open Claude Desktop → Settings → Developer → Edit Config. Restart after saving.

~/.cursor/mcp.json · .cursor/mcp.json
{
  "mcpServers": {
    "dotnet-skills-aaronontheweb": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/Aaronontheweb/dotnet-skills",
        "~/.claude/skills/dotnet"
      ],
      "_inferred": true
    }
  }
}

Cursor uses the same mcpServers schema as Claude Desktop. Project config wins over global.

VS Code → Cline → MCP Servers → Edit
{
  "mcpServers": {
    "dotnet-skills-aaronontheweb": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/Aaronontheweb/dotnet-skills",
        "~/.claude/skills/dotnet"
      ],
      "_inferred": true
    }
  }
}

Click the MCP Servers icon in the Cline sidebar, then "Edit Configuration".

~/.codeium/windsurf/mcp_config.json
{
  "mcpServers": {
    "dotnet-skills-aaronontheweb": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/Aaronontheweb/dotnet-skills",
        "~/.claude/skills/dotnet"
      ],
      "_inferred": true
    }
  }
}

Same shape as Claude Desktop. Restart Windsurf to pick up changes.

~/.continue/config.json
{
  "mcpServers": [
    {
      "name": "dotnet-skills-aaronontheweb",
      "command": "git",
      "args": [
        "clone",
        "https://github.com/Aaronontheweb/dotnet-skills",
        "~/.claude/skills/dotnet"
      ]
    }
  ]
}

Continue uses an array of server objects rather than a map.

~/.config/zed/settings.json
{
  "context_servers": {
    "dotnet-skills-aaronontheweb": {
      "command": {
        "path": "git",
        "args": [
          "clone",
          "https://github.com/Aaronontheweb/dotnet-skills",
          "~/.claude/skills/dotnet"
        ]
      }
    }
  }
}

Add to context_servers. Zed hot-reloads on save.

claude mcp add dotnet-skills-aaronontheweb -- git clone https://github.com/Aaronontheweb/dotnet-skills ~/.claude/skills/dotnet

One-liner. Verify with claude mcp list. Remove with claude mcp remove.

Use Cases

Real-world ways to use .NET Skills

Audit a service for async/await correctness

👤 Backend .NET teams ⏱ ~45 min advanced

When to use: You inherited a service that deadlocks intermittently and you suspect async issues.

Prerequisites
  • Skill installed — git clone https://github.com/Aaronontheweb/dotnet-skills ~/.claude/skills/dotnet
Flow
  1. Scan
    dotnet-skills async audit on src/. Flag: missing ConfigureAwait(false) in libraries, .Result/.Wait, async void (non-event), missing CancellationToken propagation.✓ Copied
    → Concrete file:line list with severity
  2. Fix one cluster
    Pick the highest-severity cluster (deadlock risk). Propose patches, file by file. Show diffs.✓ Copied
    → Surgical patches, not whole-file rewrites

Outcome: Async hygiene improved; one source of intermittent deadlocks gone.

Pitfalls
  • ConfigureAwait fix breaks UI thread code — Skill differentiates lib vs app code; respect that boundary
Combine with: filesystem

Benchmark a hot path with BenchmarkDotNet

👤 Perf-curious .NET devs ⏱ ~60 min advanced

When to use: Profiler points at a method; you need rigorous numbers.

Flow
  1. Scaffold benchmark
    Skill: scaffold a BenchmarkDotNet project for OrderCalc.Compute. Multiple input sizes. Memory diagnoser on. Markdown exporter.✓ Copied
    → Benchmark project compiles; runs under Release
  2. Run + interpret
    Run the bench. Report mean, allocated, gen0. What's the leading allocation?✓ Copied
    → Concrete attribution to allocator (probably string concat or LINQ)
  3. Patch + re-bench
    Replace LINQ with a hand-rolled loop. Re-run. Show before/after.✓ Copied
    → Allocations drop measurably

Outcome: Quantified perf improvement, not vibes.

Pitfalls
  • Running Debug builds — BenchmarkDotNet refuses Debug; ensure Release

Prep a library for NuGet release

👤 OSS authors ⏱ ~30 min intermediate

When to use: First NuGet release of your lib; you want it polished.

Flow
  1. Audit packaging
    dotnet-skills: audit MyLib.csproj for NuGet readiness. Check Description, Authors, License, RepositoryUrl, source-link, multi-targeting.✓ Copied
    → Concrete missing fields list
  2. API surface review
    Generate the public API surface; flag any leaked internal types or oddly-named publics.✓ Copied
    → API surface dump with concerns called out

Outcome: Clean package ready to push.

Pitfalls
  • License mismatch (LICENSE file vs <PackageLicenseExpression>) — Skill catches this; align them

Combinations

Pair with other MCPs for X10 leverage

dotnet-skills-aaronontheweb + filesystem

Apply patches across the project

Apply the async_audit patches across src/. One commit per file.✓ Copied
dotnet-skills-aaronontheweb + github

PR up the bench results

Open a PR with the perf patch + before/after benchmark in the description.✓ Copied

Tools

What this MCP exposes

ToolInputsWhen to callCost
async_audit path Codebase health check 0
scaffold_benchmark target_method Perf work 0
interpret_bench bench_md After running benchmark 0
nuget_audit csproj Pre-release 0
api_surface assembly_or_proj API review 0
test_conventions_check test_path PR review 0

Cost & Limits

What this costs to run

API quota
N/A
Tokens per call
Audits scan code = larger token use; 5k–30k per pass
Monetary
Free
Tip
Audit one project at a time; don't fan out across a monorepo at once

Security

Permissions, secrets, blast radius

Minimum scopes: filesystem-read filesystem-write
Credential storage: None
Data egress: None

Troubleshooting

Common errors and fixes

Audit too noisy

Specify severity threshold; skill defaults to all

Benchmark won't run

Must be Release build; BenchmarkDotNet refuses Debug

Verify: dotnet build -c Release
API surface includes internal types

Make sure InternalsVisibleTo is set correctly; skill reports as configured

Alternatives

.NET Skills vs others

AlternativeWhen to use it insteadTradeoff
Roslyn analyzers in repoYou want CI-time enforcementMore setup; less interactive
ReSharper / RiderYou want IDE-time fixesPaid; not LLM-driven
Aspire / .NET Cloud Native skillsCloud-native focusDifferent scope

More

Resources

📖 Read the official README on GitHub

🐙 Browse open issues

🔍 Browse all 400+ MCP servers and Skills