For Salesforce Apex teams
LATdx maps which Apex each test reaches. Change a class and the tests that touch it run on your org; tests whose inputs haven't changed return their last result.
Free for 100 tests per run · no sign-up · checksum-verified install
Parity-tested on open-source Apex projects
Each link opens the project at the commit it is pinned to. The parity suite checks LATdx's results on it against Salesforce's own run of the same tests. These are public test fixtures, not customers.
How it saves time
LATdx maps the Apex each test method statically reaches. Point it at a git diff and it runs only the methods whose reach touches a changed class.
latdx test run --affectedMetadata-only changes such as flows, validation rules and permission sets are not used for selection yet. Run the full suite when they change.
Each test method gets a fingerprint of the code and tracked metadata it reaches. When the fingerprint matches a stored result for the same org, LATdx returns that result instead of running the test.
latdx test run -o my-orgOn by default. Turn it off for one run with --no-cache.
Install
One command puts latdx on your PATH. The same CLI runs in your pipeline.
Local CLI
A single binary for your terminal. It caches Apex test results and skips reruns whose inputs did not change.
curl -fsSL https://latdx.com/install.sh | bashChecksum-verified binaryView install.shRelease notes
Then: latdx test run -o <org>
Needs a Salesforce DX project (sfdx-project.json), an authenticated sf CLI and Java 11+ on PATH. Quickstart
How it works
Every test method gets a fingerprint of the Apex it reaches and your tracked metadata. Same fingerprint, the result is reused. Different, the test runs on your org.
Your tests
AccountServiceTest
AccountTriggerTest
Apex
AccountService.cls
Trigger
Metadata
Account fields
Validation rule
AccountServiceTest
AccountTriggerTest
Step 1 of 9
LATdx parses your Apex into a graph. testHappyPath() calls validate() and save(), so that is its reach.
FAQ
What the cache does when a change sits outside what it can see.
Every test reruns. With Java 11 or newer and the LATdx daemon running (the normal setup), triggers, Flows, workflow and validation rules, profiles, permission sets and the other types listed at docs.latdx.com/test-caching fold into one shared manifest hash inside every key, so any edit, even a comment in a trigger, means a full rerun: no worse than today, plus the time LATdx spends computing keys. If Java 11+ is missing or the daemon is not running, keys fall back to one per test class and metadata edits go unseen, so re-run your suite with --force after one.
Approval processes and assignment rules fold into the shared manifest, so editing one reruns every test. Escalation and auto-response rules, queues, roles, email templates and other types missing from the docs list are not fingerprinted yet: their edits leave stored results in place and --affected selects no tests for them, so re-run the tests that depend on them with --force.
Fingerprints come from your repository, so the cache does not see an org-only change until you retrieve it, and a run made before you deploy a local edit stores the old code's result under the new fingerprint: run those tests once with --force after deploying. Results are never reused across orgs, and whenever you doubt the org, --force reruns the tests and refreshes their stored results.
A test whose key LATdx cannot compute runs on the org without storing its result, and a test class LATdx cannot parse does not run at all: it is left out with a warning. Code called by name (Type.forName, a Callable dispatch) does not invalidate stored results and --affected cannot find the tests that use it, so after changing that code re-run those tests once with --force.
Pricing
Start free with no card. Upgrade when you need unlimited tests on your machine or in CI.
Early-access rate, locked while you subscribeEarly-access rate, locked for the life of your subscription
Every plan includes method-level caching, live test results and affected-test selection.
For trying LATdx on your own suite.
$0forever
No sign-up or card required.
For developers on their own machine.
$10/ month
Billed monthly, per seat.
For teams that run Apex tests in CI.
$100/ month
5 seats included, $10 per extra seat.
Larger org or procurement process?
Talk to us about more than 250 seats, billing and your security review.
FAQ
Each Apex test method execution. The 100-test cap on the Free plan resets per run, not per day.
No. Pro unlocks unlimited tests on your machine, and CI runs stay at the 100-test Free cap until you move to Team.
The Team plan includes 5 developer seats, and extra seats are prorated to the day. Removing a seat credits the unused portion to your next invoice.
Your early-access rate stays locked for the life of your subscription. Any later price change applies only to new subscriptions.
Yes. Yearly saves about 17% (you pay for 10 months instead of 12). Switch from your billing portal.
One line installs the CLI. Then point latdx test run at your org.
curl -fsSL https://latdx.com/install.sh | bash