sf-apex

sf-apex is a Salesforce Apex code-generation and code-review skill. It helps write production-grade Apex — service classes, selectors, triggers, batch/queueable/schedulable jobs, invocable methods, and test classes — and reviews existing .cls/.trigger code against a 150-point rubric spread across eight categories (bulkification, security, testing, architecture, error handling, naming, performance, and code quality). The problem it targets is that hand-written Apex frequently hits governor limits, skips CRUD/FLS enforcement, or ships tests without assertions; the skill enforces guardrails to catch those before handoff.

Its workflow first gathers context (class type, target objects, whether code is net-new or a refactor), inspects the existing project for trigger frameworks and service/selector conventions, then generates the smallest correct pattern and scores it. It ships a template library (service, selector, batch, queueable, trigger, invocable, test-data-factory), a set of reference guides (anti-patterns, bulkification, security, testing, design patterns, SOLID, naming), and Python hook scripts that run local validation such as Apex LSP checks and Prettier formatting. Guardrails explicitly block SOQL/DML in loops, hardcoded IDs, empty catch blocks, string-built SOQL with user input, and missing sharing declarations, steering toward bind variables, WITH USER_MODE, and with-sharing defaults.

Target users are Salesforce developers, architects, and reviewers who want consistent, secure, testable Apex aligned with community best practices. The skill is security-conscious by design — much of its content is about preventing SOQL injection and enforcing field-level security — and its bundled scripts perform read-only validation and formatting rather than any destructive or network operations.