Skip to content

Awesome Humans of .NET

An agentic take on awesome-dotnet: opinionated guidance for your coding agent, distilled from the writing of awesome humans. Instead of a list of links, it gives one recommendation per question, and each opinion cites its sources. It documents current best practice in .NET code, and changes as .NET evolves.

The writers are listed in AWESOME-HUMANS.md: people and publications with at least two years of accurate, independent writing on .NET.

Agents: Fetch llms.txt and every file it links as raw text. It links the opinions, templates, and skills.

Usage

  • Point your agent at it. Tell your coding agent to follow the opinions when it writes code:

    Follow the conventions at awesome-humans.net/dotnet/llms.txt when writing .NET code
    

    The opinions cover conventions, project layout, language usage, and library choices.

  • Check an existing project. Ask your coding agent to run review-project:

    Use the review-project skill at awesome-humans.net/dotnet/llms.txt on <repository>
    

    To narrow the review, say what to include or leave out, and optionally name a house profile:

    Use the review-project skill at awesome-humans.net/dotnet/llms.txt on <repository>, focus on logging and testing, skip CI, use the simon-mckenna house profile
    
  • Research a topic. Ask your coding agent to run research-topic, which answers from the opinions and the vetted sources, citing each claim:

    Use the research-topic skill at awesome-humans.net/dotnet/llms.txt to research whether to turn on runtime async
    
  • Read the opinions and templates. Browse them at awesome-humans.net or on GitHub. Each opinion gives its recommendation, rationale, and sources, and the templates/ encode them.

Scope

opinions/ holds one file per topic:

  • Application architecture: modular monoliths, vertical slices, and when layering is worth its cost
  • ASP.NET Core: minimal APIs, hosting, request validation, error responses, OpenAPI, and performance
  • ASP.NET Core security: authentication, authorization, browser sessions and BFF, tokens, CSRF, CORS, and security headers
  • C#: the latest released language version and idiomatic use of its features
  • CI & automation: supply-chain security, pipeline stages, package signing, and dependency updates
  • Data access: EF Core set-based writes, date and time mapping, PostgreSQL indexes, raw SQL, and tenant isolation with row-level security
  • Dates, times, and time zones: choosing a type, UTC versus local storage, TimeProvider, and testing time
  • F#: domain modelling, mixed C# and F# solutions, and testing
  • Globalization & localization: culture versus ordinal comparison, ICU and invariant mode, IStringLocalizer, and the data that containers drop
  • Logging & tracing: structured logging, source-generated log messages, and OpenTelemetry over OTLP
  • Project structure & SDK: project files, solution formats, SDK pinning, Central Package Management, repository layout, .editorconfig, and containers
  • Runtime & performance: measuring first, immutable collections, async code, Span<T> and pooling, the GC, and Native AOT
  • Security: secrets, deserialization, cryptography, and servicing cadence
  • Testing: choosing a framework, integration tests, deterministic time, coverage, and proving that a test can fail
  • UI frameworks: Blazor and WebAssembly, .NET MAUI, and cross-platform desktop with Avalonia
  • Libraries: what to use and what to avoid, covered in the files above

Freshness policy

Opinions target the latest released .NET, C#, and F#, never an older LTS release. Preview features appear only in "Coming next" notes. The skills move the repository to each new version.

Every resource records when it was last reviewed against its sources. agents/metadata.md defines the fields, and CI enforces them.

Awesome humans

Every opinion cites a vetted source: a person, such as Stephen Toub or Andrew Lock, or a publication, such as the .NET Blog or Microsoft Learn. A source is admitted on its track record: at least two years of sustained writing, with depth, accuracy, and an independent view. Videos, talks, and podcasts are out of scope for now, because an opinion cites text a reader can check. AWESOME-HUMANS.md has the roster and the criteria.

This diagram shows how a source is admitted and what its standing allows. The admission criteria in AWESOME-HUMANS.md and the vet-source skill are canonical.

Source standing: vet-source sorts a candidate into citable, watch list or declined, and a citable source is unmarked, Corroborate or Discovery-only

Standing can change. vet-source runs again when an admitted source goes dormant or drops in quality, and when a watch-listed source's blocker clears.

Two markings in a source's notes limit how it's cited:

  • Corroborate: A source marked **Corroborate.** is never the only citation for a claim.
  • Discovery-only: A source marked **Discovery-only.** is never cited. Cite the primary source it links to instead.

Any other limit, such as a concern about independence or an outdated back catalogue, goes in the source's notes as prose, and doesn't change its standing.

House opinions

A house opinion is a preference from experience, not from a vetted source. Each belongs to a house profile, named for the person or organisation that owns it, such as simon-mckenna. On awesome-humans.net, every page shows only sourced community practice until you choose a profile in the header. An agent that follows a profile reads llms-<profile>.txt instead of llms.txt.

In this repository, each house opinion is a marked block beside the sourced text it changes, and HOUSE-OPINIONS.md defines the format and records each one. Other contributors can propose sourced or experience-based opinions through the organisation's opinion pull request template.