Complete AI Training
Sign inGet my AI kit

Your job's AI kit

Get your AI kit

Tell us who you are and what you do. We show you your kit right away and email you the link: skills, prompts, AI agents, MCP servers and courses for your job.

500+ jobs ready, and we make a kit for any other job. No payment needed to look.

Share

AI agent for qa managers

Performance Regression Profiling Agent

Catch performance regressions the day they land

MonitoringCreativesIT and developmentProduct development
See how it worksGet access
Performance Regression Profiling Agent: what goes in, what the agent does and what you get

What it does

Performance slowly gets worse when nobody notices that one change added a few milliseconds per frame. Every night this agent runs the game's performance test routes on target hardware, recording frame time, memory, load times and hitches. It compares each result with the last good build. When a metric worsens past the threshold, it repeats the run three times to rule out noise. A confirmed regression starts a bisect: it builds and tests the day's changes in halves to find the one that caused it. It then confirms the cause by testing the builds just before and after that change. If they do not confirm it, it keeps bisecting or tests combinations of changes. It drafts a ticket for the change author with profiler captures. The lead approves priority. Edge case: background process noise is handled by repeat runs before anything is reported.

How it works

Follow the arrows from top to bottom. The orange dashed arrow is the loop: when a check fails, the agent goes back and tries again.

Start and resultWhat it doesA check on its own workWaits for your OKGoes back and retries
Yes, continueYes, continueApprovedNoNo 1 STARTS WHEN Nightly build ready 2 USES A TOOL Run performance routes on target hardware 3 DOES Compare with the last good build 4 CHECKS THE RESULT Is the regression confirmed on repeat runs? If not: mark as noise and continue. Back to step 2. 5 USES A TOOL Bisect the day's changes 6 CHECKS THE RESULT Do builds before and after the change confirm thecause? If not: keep bisecting or test change combinations. Backto step 5. 7 USES A TOOL Draft ticket with captures 8 YOU APPROVE Lead approves priority 9 RESULT Regression ticket filed
Read the steps as a list
  1. Nightly build ready
  2. Run performance routes on target hardware
  3. Compare with the last good build
  4. Is the regression confirmed on repeat runs?If not: mark as noise and continue. Back to step 2.
  5. Bisect the day's changes
  6. Do builds before and after the change confirm the cause?If not: keep bisecting or test change combinations. Back to step 5.
  7. Draft ticket with captures
  8. Lead approves priorityThe agent waits here for your OK.
  9. Regression ticket filed

How it decides

It declares a regression only when repeated runs exceed the threshold, then bisects to the change.

  • Threshold 5% frame time or 50 MB memory
  • Three repeat runs before declaring
  • Assign to the change author

Make it yours

Every agent is a starting point. You choose these settings for your own situation.

  • Thresholds
  • Routes and hardware
  • Repeat count
  • Ticket template

What keeps you in control

It always asks you first

  • Priority
  • Reverting changes

Hard limits

  • Never reverts changes itself
  • Never changes budgets

It stops when

  • Done: ticket filed
  • Stop: build broken

Set it up

We guide you through the set-up, step by step

Members get the full set-up guide for this agent. No technical skills needed: you copy, paste and upload.

10 minto set it up in your AI
5 AIsChatGPT, Claude, Copilot, Gemini, Grok
  • One set of instructions to paste into your AI, with the clicks for ChatGPT, Claude, Microsoft 365 Copilot, Gemini and Grok
  • The agent then walks you through connecting your own data, one source at a time
  • A downloadable copy with the flow chart, the rules and the full guide
Get access to this agent

An example run

What happensOn the October 14 nightly, frame time on the city route rose from 15.8 to 17.4 ms, past the 5% threshold. Three repeat runs confirmed it. Bisecting 23 changes first pointed to a lighting tweak, but the before-and-after check did not confirm it. The agent kept bisecting and found a new particle effect, confirmed by both builds. The lead approved high priority for the author's ticket.

More agents for qa managers

Quality control · Coding

Public Form Accessibility Agent

Every public form remains accessible after each revision.

7 steps2 checks5 jobs
Coding · Automation

Flaky Test Reproduction Agent

Every flaky test ends with a reproducible cause or a documented set of ruled-out causes.

9 steps1 check3 jobs
Coding · IT and development

Accessibility Regression Repair Agent

Keyboard and screen-reader journeys keep working after UI changes.

7 steps2 checks6 jobs
Coding · IT and development

Test Fixture Data Minimization Agent

Fixtures with no unnecessary personal data and all tests passing.

7 steps1 check2 jobs