dotfiles

Load when designing a new API endpoint, choosing HTTP methods or status codes, designing request/response schemas, deciding on URL structure, handling pagination, or reviewing an endpoint for REST consistency. Ensures every endpoint in the API follows the same conventions so the API is predictable and easy to consume.

Load when writing, reviewing, or debugging C# code — ASP.NET Core APIs, background services, console apps, or libraries. Covers the patterns Claude consistently gets wrong in production C#: async/await pitfalls, nullable reference types, DI patterns, cancellation tokens, and LINQ vs. loops. Do not load for Python, Go, or Rust.

Load when investigating a bug, traceback, unexpected behavior, test failure, or performance anomaly in Python or React. Also load when asked to explain an error or trace through why something isn't working as expected. Provides a systematic methodology — not just pattern-matching on error messages.

Load when writing, reviewing, or debugging Django code — models, views, DRF serializers, viewsets, URL routing, signals, or admin. Covers the patterns Claude consistently gets wrong: queryset filtering, N+1 via ORM relationships, DRF serializer validation, permission classes, and queryset ownership scoping. Load alongside testing-django when writing tests.

Load when writing, reviewing, or debugging FastAPI code — route handlers, dependencies, background tasks, middleware, or lifespan events. Covers the patterns Claude consistently gets wrong: business logic in handlers, dependency injection for sessions, response models, status codes, and async pitfalls. Load alongside sqlalchemy-postgres when DB access is involved.

Load when writing a commit message, creating a PR description, naming a branch, reviewing git history, or squashing commits. Also load when asked to summarize changes for a changelog or release note. Keeps commit history clean, semantic, and useful as a debugging artifact.

Load when writing, reviewing, or debugging Go code — services, CLI tools, HTTP handlers, concurrency patterns, or tests. Covers the patterns Claude consistently gets wrong in production Go: error wrapping, goroutine leaks, context propagation, and project layout. Do not load for Python or other languages.

Load when adding logging to application code, setting up structured logging, configuring Sentry error tracking, designing log levels, or reviewing code for observability gaps. Also load when debugging production issues where log output is part of the investigation. Keeps logs useful, consistent, and safe (no sensitive data).

Load when writing, reviewing, or debugging general Python code — not framework- specific. Covers the patterns Claude consistently gets wrong in production Python: mutable defaults, exception handling, type hints, dataclasses, closures, and common gotchas. Load this alongside a framework skill (fastapi, django) when relevant.

Load when writing, reviewing, or debugging React/TypeScript code — components, hooks, data fetching with TanStack Query, forms, and state management. Covers the patterns Claude consistently gets wrong: useEffect for data fetching, missing loading/error states, Zod schema mismatches, and TypeScript anti-patterns.

Load when writing, reviewing, or debugging Rust code — services, CLI tools, async code with Tokio, or systems programming. Covers the patterns Claude consistently gets wrong in production Rust: unnecessary clones, error handling with thiserror/anyhow, async pitfalls, and lifetime over-engineering. Do not load for Python, Go, or other languages.

Load when reviewing code for security issues, writing authentication or authorization logic, or scaffolding new endpoints. Provides a systematic checklist covering injection, authentication, authorization, data exposure, secrets, and input validation. Language-agnostic — applies to Python, Go, Rust, C#.

Load when writing, reviewing, or debugging SQLAlchemy 2.x ORM code against PostgreSQL. Covers the patterns Claude consistently gets wrong: select() vs query(), async session usage, relationship loading, N+1 queries, transactions, and Alembic migration gotchas.

Load when writing or reviewing tests for Django or Django REST Framework — APITestCase, pytest-django, factory_boy, authentication in tests, and permission testing. Covers the patterns Claude consistently gets wrong: force_authenticate vs token auth, missing django_db marks, queryset assertions, and incomplete permission coverage.

Load when writing or reviewing Python tests — pytest fixtures, factory_boy, async tests, mocking, and test structure. Covers the patterns Claude consistently gets wrong: over-mocking, fixture scope, test naming, and the difference between unit and integration tests.