Skip to main content
You never name any of this explicitly. You say something like the example in each section, and your agent works out what to do. The lists are what it is able to do for you, not a menu you type. Before any of it, your agent works out which Peeps project you mean — either from your checkout’s git remote, or by listing what you have access to. You rarely need to tell it.
Peeps already does most of this on its own: it runs the suite on every change, analyzes each failure, heals script issues, and re-verifies before publishing. What follows is what you can reach on demand, with your local code in the same conversation.

​
Understand failures

Why is the checkout test failing?
  • See everything currently failing in a project
  • Open one failure in full: cause and confidence, the exact error with line context, the script that ran, the failure screenshot, run history, and what it depends on
  • Summarise a whole batch run
  • Read a specific test case, or search across them
  • Pull a single artifact from a run
  • Show the execution plan behind a run

​
Run tests

Run the checkout suite against staging.
  • Kick off a test run in Peeps’s hosted CI
  • Check where that run has got to
Run plans identify production targets and queue nothing by themselves. Your agent shows you the plan and waits for your approval before anything runs.

​
Fix tests

Fix that selector and submit it.
  • Submit a proposed fix to a test script
  • Check whether that fix verified
  • Update the written steps of a test case
Peeps self-heals script issues on its own, so this path is for what healing hands back: a fix that needs your local code to get right, or a failure Peeps diagnosed as something other than a broken selector. A submitted fix is verified by Peeps in its own hosted run. Promoting a verified fix is a person’s job in the Peeps web app, not your agent’s.

​
Create tests

Do we have Peeps tests for password reset?
  • Create new test cases
  • Generate and publish a script for a test case
  • Check how that generation is going

Creating tests, walked through

Four ways in, one engine — coverage checks, branch diffs, single gaps, and converting existing Playwright tests.