Salesforce Test Runner

Run Apex tests from a sidebar built for the job: pick the target org, tick classes and methods across your workspace and the org, read parsed results with clickable failures, and see coverage in its own view and painted on your editor lines. Every sf CLI command we execute is listed in a collapsible log. Reuses your existing sf CLI auth — no separate OAuth flow.

Features

The SF Tests activity-bar icon opens four views. Each is a collapsible section: fold the ones you are not using, resize them, or hide them from the container's ⋯ menu.

• Tests — the target org as a dropdown, ⟳ to refresh the org list and ＋ to log in to another org; a production org gets a warning bar above the tree. All/Selected tabs, a search box and a source filter sit above a tree of test classes with tri-state checkboxes down to the method. Select tests for active class ticks the tests for the class open in the editor, and a “with coverage” checkbox decides whether the run asks for --code-coverage. A “with logs” checkbox keeps each test method's debug log: before the run the plugin makes sure your user has an active USER_DEBUG trace flag (extending yours, creating one on an SfTestRunner debug level that logs Apex at DEBUG and nothing else, or pointing yours at that level when its own would swallow System.debug), and afterwards a log button next to each method opens its log in an editor tab — the System.debug lines first, the full log underneath, ready to read, copy or save. The run row is Run Selected, All Local (--test-level RunLocalTests, every test in the org except managed-package ones) and All in Org (managed packages included). While a run is on you see done/total, the elapsed time and a Cancel button.
• Tests from the org, not only from disk — Fetch org tests merges the org's test classes with the workspace scan. Classes only in the org are listed as org-only and are runnable; classes only on disk carry a not deployed badge and a warning before a run. The org's list is cached per org and stamped "as of".
• Results — one run bar (PASS/FAIL, counts, wall time, org, finish time, Re-run failed, Copy), an All/Failed filter, and a tree of test classes with n/m passed and time, each method with its outcome and duration. A failure expands in place with the assertion message and every parsed stack frame as a link that opens the file at that line; compile failures get their own badge. The tree fills in as methods finish.
• Coverage — after a coverage run the view leads with the classes the run was aimed at, not everything it touched. Targets are resolved per test class: Declared from @IsTest(testFor='ApexClass:Foo') (API v66+, and it names triggers too), else By name from the FooTest/TestFoo/Foo_Test/FooTests conventions when that class really was covered, else By truncated name for a class too long to have a conventional test class — matched on a prefix, declining on a tie. The headline and the bar measure those classes, with the 75% deploy floor dimmed to say it is an org-wide rule; the whole-run average sits in the legend. A declared target the run never reached shows as not exercised, and everything else folds into Also covered. A row opens the class or trigger; its cloud button loads that one class's stored coverage from the org's last run, whoever ran it, labelled as such. Covered and uncovered lines are painted in the editor with a gutter bar and overview-ruler marks, the run and org are named in the hover, and a status-bar eye shows the active file's percentage and toggles the painting. Coverage only ever comes from your own run or that explicit per-class load; it dims once you edit the file and is dropped when you switch orgs.
• CodeLens on test classes and methods — ▶ Run and Run with Coverage, plus the last outcome.
• Cancel aborts in the org — queued classes are marked Aborted; the class already executing finishes.
• Load Recent Test Runs — pull in any recent async run of the org, including runs started from a terminal, CI, or lost to a window reload; it lands in the Results view like your own.
• Production runs ask first — a run against an org classified as production needs a modal confirmation before anything starts.
• Command log — every CLI invocation with its full command, duration and any error; Copy Command drops the exact invocation into your clipboard. Starts collapsed.
• Your org is your own — the choice made here is remembered by this extension, per VS Code window (each open workspace keeps its own target org), and does not follow the org other tools select. Set sfTestRunner.syncOrgWithFamily to true to follow (and publish) the org shared with the other Skrety Salesforce extensions.

Requirements

• VS Code 1.88 or newer.
• The Salesforce CLI (sf) installed and on PATH.
• An authenticated org (sf org login web) — typically the same auth your other Salesforce VS Code extensions use.

Commands

Settings

How it talks to Salesforce

Every operation flows through one wrapper around the sf CLI. The exact commands you'll see in the log:

• Starting a run — sf apex run test --tests <ClassName|Class.method> [--tests …] [--code-coverage] --result-format json --json --target-org <username> (no --wait: the run id comes back at once). Whole-suite runs use --test-level RunLocalTests or RunAllTestsInOrg instead of --tests.
• Progress — sf data query --use-tooling-api --json on ApexTestRunResult (status and method counts) and ApexTestResult (finished methods) every few seconds.
• Results and coverage — sf apex get test --test-run-id <id> --code-coverage --result-format json --target-org <username> once the run is done; the same command loads a recent run.
• Cancel — the run's ApexTestQueueItem rows still queued are set to Aborted with one sf api request rest -X PATCH call per 200 rows.
• Org test classes — sf data query --use-tooling-api for ApexClass names, then the source of classes not in the workspace, in batches, to find the ones with tests.
• Coverage from org — sf data query --query "SELECT … FROM ApexCodeCoverageAggregate WHERE …" --use-tooling-api --json.
• Org listing — sf org list --skip-connection-status --json; login — sf org login web --json.

Because we never embed your access token directly, your sf CLI auth is the single source of truth. Re-auth with sf org login web and the extension picks it up immediately.

Known limitations

• Workspace discovery is a regex scan of your .cls sources, not an Apex parser. A test the CLI ran but the scan did not find still appears in the Results view under its class. A class annotated @IsTest with no test methods in it — a data factory, a callout mock — is deliberately not listed, since running it could only fail.
• Org discovery reads class source in batches; on an org with thousands of classes not present in your workspace the first fetch takes a while. It is cached per org afterwards.
• A run only reports coverage for the classes it exercised — never for the test class itself, which is how Salesforce reports it. So after running FooTest you see the coverage in Foo.
• Classes only: triggers are not discovered as tests, and a trigger's coverage is shown in the table but not painted.
• Cancelling stops what is still queued; the class the org is executing at that moment finishes.
• Only runs started by this extension appear by themselves; use Load Recent Test Runs for anything else.

License

MIT — see the bundled LICENSE file.