platform-apex-test-run

platform-apex-test-run handles Salesforce Apex test execution, coverage analysis, and disciplined test-fix loops, with a 120-point scoring guide for judging test confidence. It solves the problem of running Apex tests, checking code coverage, interpreting failures, improving coverage, and managing structured fix loops for Salesforce code. It owns tasks involving sf apex run test workflows, Apex unit-test failures, coverage analysis, identifying uncovered lines and missing scenarios, and structured test-fix loops, while delegating production code work, Agentforce agent testing, and LWC/Jest testing to other skills.

The recommended workflow gathers context first (target org alias, test scope, coverage threshold, diagnosis-only vs. fix loop, existing data factories), then discovers test scope, runs the smallest useful test set first, analyzes results (failing methods, exception types and stack traces, uncovered lines), runs a disciplined fix loop, and improves coverage intentionally across positive, negative, bulk (251+ records), and callout/async paths. It ships high-signal rules such as defaulting to SeeAllData=false, requiring meaningful assertions, testing bulk behavior with 251+ records to cross the 200-record trigger batch boundary, using @TestSetup, and pairing Test.startTest() with Test.stopTest() for async. A rich set of reference files and asset templates is included — for example references/cli-commands.md, references/mocking-patterns.md, assets/basic-test.cls, assets/bulk-test.cls, assets/dml-mock.cls, and scripts/parse-test-results.py, a post-tool hook that parses sf apex run test JSON output for the auto-fix loop.

Target users are Salesforce developers running Apex tests, checking coverage, or fixing failing tests, especially when touching *Test.cls / *_Test.cls files. It documents common gotchas (tests passing locally but failing in CI, coverage drops after refactor, uncommitted-work callout errors, mocks not taking effect, @TestSetup data requiring re-query) and a fixed output format reporting tests run, pass/fail, coverage, root-cause findings, and next steps. Its Score Guide interprets results, with 108+ indicating strong production-grade confidence and below 84 meaning below standard.