Summary

Writing a test becomes writing a prompt, and the loop from there to a red-or-green report never leaves the conversation.

Three separate generation paths — a single discovery, a batch from one prompt and images, a set built from a written test plan — are the reason to reach for this rather than the dashboard, and executeTests with getTestReport closes the loop in the same session. Repairs are finer-grained than authoring: updateTestCaseElement changes one element's locator line at a time, so a broken selector is quick and a restructured page is not.

What it is

A server over Octomind's end-to-end testing platform, 20 tools spanning test targets, environments, generated test cases, executions and reports.

What you get

Test targets — the applications under test — created, listed, updated and deleted, with settings such as testIdAttribute and timeoutPerStep
Environments per target holding a discovery URL, a test account, header configuration, basic auth and variables, created, updated and deleted
Test cases generated from a prompt by discovery, many at once by createBatchGeneration with optional images, or from freeform plan text by createFromTestPlan
executeTests triggering a run against a URL for a chosen environment, with tags and per-run variable overrides
Reports listed and fetched per target, and the individual test case behind a result retrieved with its interactions and assertions
Existing cases edited after the fact — description, folder, tags, run status — and updateTestCaseElement reaching a single step's locator line
Private locations listed, which is how a target behind a firewall or VPN gets reached

Requirements

An Octomind account and its key in APIKEY. Tests run against a URL you supply, so the deployment has to be reachable from Octomind — for one that is not, an environment needs a private location configured for your organisation.

Setup effort

One command plus a key — npx @octomind/octomind-mcp --clients, then supply credentials