> ## Documentation Index > Fetch the complete documentation index at: https://apidocs.neetoplaydash.com/llms.txt > Use this file to discover all available pages before exploring further. # Examples > Prompts that show what the connection can answer. You do not need to name tools or remember arguments. Describe what you want and the assistant picks the right calls. These prompts are grouped by the job they do. ## Find your way around > List my NeetoPlaydash projects. > Show the last 10 runs for Spinkart Web on the main branch. > Which of last night's runs failed? The assistant starts from `ListProjects`, then filters runs by branch and status. ## Triage a red run > What went wrong in the latest failed run on main? > Group the failures in this run by their error and show me the biggest cause first. > Which spec files have failing tests in this run? The second prompt is the one worth learning. It maps to `GetTopErrors`, which collapses failing tests onto the error they share - so a run with forty failures usually resolves to two or three real problems. ## Dig into one test > Show me the error output for the checkout test that failed. > How many attempts did that test take, and what did each one do? > Get me the Playwright trace for that failure. Traces open in Playwright's hosted viewer, where you can step through the DOM, network calls, and console output at the moment things broke. ## Tell a flake from a break > Has this test been failing recently, or is this new? > Show me this test's history over the last 30 days. > Which tests in this run are flaky rather than genuinely broken? Result history is what separates "this broke today" from "this has been red on and off for a fortnight", which usually decides whether you fix the test or the code. ## Watch the trend > How has the pass rate on main moved over the last month? > Show me the daily flaky test count for the nightly runs since 15 July. > Which day in the last 30 had the worst failure rate, and how long were runs taking that day? These map to `GetProjectInsights`, which aggregates by day rather than reading individual runs. It covers the most recent three months. ## Work across workspaces > Which NeetoPlaydash workspaces can you see? > List the failing runs in the acme workspace. An OAuth connection can read every workspace you approved during sign-in. An API key connection has exactly one. ## Put it together > Summarize last night's run on main: how many tests failed, what were the top errors, and which of those failures are new rather than flaky? This is the prompt the whole connection exists for. It fans out across runs, top errors, and result history, and comes back with the shape of the problem rather than a wall of logs. Every tool is read-only, so none of these prompts can change your workspace. See [Tools](/mcp/tools) for the exact list.