agentleFS
LeanMeanCleanAppsLeanMeanCleanApps/CleanGraphQLApi/.agents/skills/api-conventions/SKILL.md

api-conventions

Skill13 starsChanged 10 days agoMITView on GitHub

How this GraphQL schema is named and shaped - camelCase fields, nullability, Relay connections over offset paging with opaque cursors, mutation conventions with typed ValidationError and NotFoundError payload errors, the VALIDATION and NOT_FOUND error codes on queries, depth limits and authorization. Use when naming a field or argument, deciding nullability, returning a list, or changing how errors appear.

15 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

3 rules. 0 are common, 2 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 11 edited versions we read keep these.

  • 2

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Test each error code or error type in the field's integration tests (test-design).

    line 48
  2. Distinctive:

    Put stack traces or internal detail in an error message or extension.

    line 64
  3. Distinctive:

    Add a second error filter.

    line 65

“Common” is measured: kept by at least 6 in 10 of the 11 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

API conventions (GraphQL)

The rule

ThingConvention
Fields and argumentscamelCase, named for what they return: movie(id), movies(query, first, after)
TypesThe Application entity's name: Movie, MovieSummary, Review
MutationsHot Chocolate's mutation conventions: createReview(input: CreateReviewInput!): CreateReviewPayload!, where the payload holds the result (review) and errors
IdsUUID
NullabilityEntity fields are non-null. A field that can fail returns a nullable result, so an error nulls only that field
ListsRelay connections (nodes, edges, pageInfo, totalCount), forward only, first defaults to 10 and is at most 50
DepthQueries deeper than 15 levels are refused (AddMaxExecutionDepthRule(15))
CostHot Chocolate's cost analysis is on, so the schema carries @cost and @listSize and expensive queries are refused

Connections sit on top of the shared offset paging. PageExtensions in GraphQL/Extensions/ makes each cursor a base64 offset:<n> and turns an after cursor back into an offset. A cursor that didn't come from us is a validation error.

Mutations report the failures a caller can fix as typed errors in the payload, through [Error<T>] on the mutation. Each error type in GraphQL/Errors/ takes the Application exception in its constructor and implements the Error interface, so it has a message:

ExceptionError typeFields
ValidationExceptionValidationErrormessage "One or more validation errors occurred.", and fields, each with field and messages
NotFoundExceptionNotFoundErrormessage, the exception's message

Queries, and anything a mutation doesn't declare, go through ApplicationErrorFilter in GraphQL/Errors/, which maps Application exceptions onto GraphQL errors:

Exceptionextensions.codeMessage and extensions
ValidationExceptionVALIDATION"One or more validation errors occurred.", with extensions.errors grouping messages by property
NotFoundExceptionNOT_FOUNDThe exception's message
Anything elseHot Chocolate's own"Unexpected Execution Error", with the exception details hidden unless a debugger is attached

Hot Chocolate returns an unexpected exception to the client and logs nothing, so the filter logs it as Error, with the exception and the field path (serilog). The filter is built from the schema's services, so ProgramExtensions hands it a logger from the application's with GetRootServiceProvider().

Steps

  1. Name new fields and arguments after the data, not the Application class (reviews, not getReviewsQuery).
  2. Decide nullability on purpose: non-null for fields that always have a value, nullable for a field whose resolver can throw.
  3. Return a connection for any list that can grow. Never an unbounded list.
  4. Report a new kind of failure as an Application exception. Add an error type for it in GraphQL/Errors/ and an [Error<T>] on each mutation that can raise it, and a case to ApplicationErrorFilter with its own code for queries, with a unit test in tests/Presentation.Tests.Unit/GraphQL/Errors/.
  5. Test each error code or error type in the field's integration tests (test-design). Ask for errors with errors{__typename,...on Error{message}}.

Example

// src/Presentation/GraphQL/Errors/ApplicationErrorFilter.cs
NotFoundException notFoundException => ErrorBuilder.FromError(error)
    .SetMessage(notFoundException.Message)
    .SetCode("NOT_FOUND")
    .Build(),

Don't

  • Expose offset arguments or backward paging.
  • Change a field's name, type or nullability without updating the schema snapshot and saying so in the pull request. It breaks clients.
  • Put stack traces or internal detail in an error message or extension.
  • Add a second error filter.

Check

dotnet test tests/Presentation.Tests.Integration

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.