When /ultrareview earns its cloud-run cost

/ultrareview launches a multi-agent cloud review of the current branch. Different subagents look at security, correctness, style, and test coverage in parallel and reconcile their findings into a single punch list. Running it costs compute — your plan covers a few free runs, but over-using it in CI will surprise your billing.

When it's worth it:

Before merging a branch that touches auth, payments, PII, or external APIs.
Before a refactor PR where "does it still work" is load-bearing.
When a teammate asks "is this good enough to ship?" and you want a second read in under five minutes.

When it's not:

Bug fixes with tests. The tests are the review.
Docs changes, renames, or bumps. There's nothing for the security or correctness agents to find.
Exploratory branches you're not shipping. You can always run it later.

Tip: pair with /ultraplan for multi-week work — the planner produces the approach, you implement, the reviewer grades the result against the plan. Three agents, one cohesive loop.

If /ultrareview says "ship it," that's meaningful. If it flags an issue, fix it or document why it's wrong. The worst thing you can do is run it and ignore the output — then you're paying for second opinions you're not listening to.

When it's worth it:

Before merging a branch that touches auth, payments, PII, or external APIs.
Before a refactor PR where "does it still work" is load-bearing.
When a teammate asks "is this good enough to ship?" and you want a second read in under five minutes.

When it's not:

Bug fixes with tests. The tests are the review.
Docs changes, renames, or bumps. There's nothing for the security or correctness agents to find.
Exploratory branches you're not shipping. You can always run it later.

Tip: pair with /ultraplan for multi-week work — the planner produces the approach, you implement, the reviewer grades the result against the plan. Three agents, one cohesive loop.