A QA agent separate from the build agent
Agencies and freelancersShop that keeps building and testing apart
The QA agent tested features against old requirements because it didn't know what had changed.
How does a separate QA agent know what the build agent changed and why?
Run both agents in the same ArcRoom project. The build agent saves what it changed and why, and the QA agent gets the change and the decision behind it, so it tests against current requirements. Bugs the QA agent finds are saved for the build agent's next session.
Also asked as: Our QA agent keeps testing against old requirements. How do we fix that? · How do I pass changes from a build agent to a testing agent? · Can a testing agent see the decisions behind a feature change?
The build agent saves what it changed and why.
- Decision
[QA] Changed today: checkout now allows guest users (decision: Oct 30). Testing guest flow.
Step by step
- The build agent saves what it changed and why.
- The QA agent gets the change and the decision behind it.
- Bugs it finds are saved for the build agent's next session.
What the agent sees
[QA] Changed today: checkout now allows guest users (decision: Oct 30). Testing guest flow.Compared with today's tools
Ticket acceptance criteria (Jira or Linear)
The team updates the ticket when requirements change and points the QA agent at the ticket text.
- Setup
- already in place for most teams
- Ongoing
- keep tickets current and paste them in, ~5 min per feature
ArcRoom
Every agent reads the team playbook before it works and writes to it when it finishes. Nothing to update by hand.
- Setup
- 2 min per project
- Ongoing
- None
Tickets are the right home for requirements; ArcRoom adds the build agent's own record of what changed today, which often isn't in the ticket yet.
Where it stops
Both agents must run in the same project.
Try it in your project
Works with the agents your team already uses, and any agent over MCP or the HTTP API.