MAXIMUX 0
Check your own services under load — safely.
You choose a service you own, you approve it, and the system runs a bounded test and keeps a record. Nothing runs against a service nobody approved.
Words you will see
Plain meanings for the terms used across the app.
- Target
- A service you own, such as
api.example.com. It must be approved before anything can run against it. - Approval
- A person's yes, with a short note explaining why the test is allowed.
- Scope
- The hosts a target covers. A test can only reach hosts inside its scope.
- Test
- A bounded run that sends requests to a target for a set time at a set rate.
- Kill switch
- One switch that stops all testing for your organization right away.
- Rate cap
- A ceiling on how many requests per second all running tests may send together.
- Agent
- An assistant that reads your request, uses the app's tools as you, and reports back.
- Workflow
- A saved list of steps the agent runs in order, such as a health check.
- Proposal
- A code change the agent suggests. Nothing changes until a person applies it.
- Audit trail
- The record of who did what and when. Each entry is linked to the one before it, so edits show up.
How a test moves through the system
Each step either moves forward or stops with a clear reason.
- Add a target. Name the service and the hosts it uses.
- Get an approval. A person writes why the test is allowed. Without it, nothing runs.
- Start a test. Set how long it runs, how many users it imitates, and the request rate. Limits are checked for you.
- Watch it. The worker reports progress. You can stop it at any time.
Working with the agent
Type or speak a goal in the chat. The agent shows each step as it happens.
When a step fails, the agent reads the error, fixes its request, and tries again. A fix that works is saved as a lesson for next time.
How much access does each role have?
A relative picture, not data. Higher bars can do more.
Only owners can switch agent codebase edit on or off. Admins and owners can read the audit trail.
Staying in control
- Only approved targets can be tested. Approval needs a written reason.
- One switch stops everything. The kill switch cancels queued and running tests.
- The agent proposes, people apply. Code changes wait for a person to approve them.
- Every action is recorded. The audit trail shows who did what, and it can be checked for edits.
Need help?
Ask in the chat, or type help in the CLI tab. An admin can read the audit trail to see what happened and when.