Reviewing code and driving git in natural language
Local code review through /code-review offers five effort levels from low to max, and ultra is not a sixth level of the same scale but a distinct facility, a multi agent review in the cloud billed separately.
A local code review is launched with /code-review and five effort levels, low, medium, high, xhigh, max. This is not a continuous scale that goes up to a so called ultra review: ultra is a distinct facility, an escalation to a multi agent review run in the cloud, billed separately from the five local levels included in normal plan usage.
Five local levels, one confidence report
The low and medium levels only report findings the review is most confident about, while high, xhigh, and max widen the coverage at the cost of including less certain findings. Choosing the effort proportionate to the risk of the change avoids two symmetric pitfalls: too low an effort that lets a real bug through, too high an effort that drowns a small fix under uncertain remarks. The level is typed straight after the command, with no flag in front of it, everything that follows /code-review without the word ultra being read as the target of the review rather than as an option.
/code-review medium
# rapporte les constats les plus sûrs sur le diff en cours
# /code-review max élargirait la couverture,
# au prix d'inclure des constats moins certainsUltra is not a sixth level
The ultra command triggers a multi agent review in the cloud, distinct from the five local levels and billed independently: a Pro or Max account is granted three free runs allocated once, none on Team or Enterprise, beyond that the cost typically ranges from five to twenty five dollars in usage credits depending on the size of the change analysed, with confirmation requested before every interactive launch. A third facility exists as well, reserved for Team and Enterprise plans: a GitHub app service that automatically comments on every pull request, billed separately at an average of fifteen to twenty five dollars per review.
Git is driven without memorising commands
Asking in plain language for the repository status, the commit message, or the creation of a pull request actually triggers the underlying commands, gh pr create or glab mr create depending on the platform, without the user needing to know their exact syntax. The safety net stays the same as with these commands typed by hand: forcing a push or deleting a branch remains subject to the permission mode chosen for the session, exactly like any other git command. The review effort chosen above applies before this git step, just as a regression test precedes the closing of a bug.
Three review facilities, not a single scale
| Three ways to have code reviewed | Where the review runs | Cost | What it covers |
|---|---|---|---|
| /code-review, low to max levels | Locally, within the current session | Included in normal plan usage | A single agent, five effort levels from most certain to widest |
| ultra, multi agent review | In the cloud | Three free runs on Pro and Max, none on Team or Enterprise, then typically five to twenty five dollars | Several distinct agents on the same diff |
| Code Review, GitHub app | On GitHub, on every pull request | Fifteen to twenty five dollars per review on average, reserved for Team and Enterprise | Automatic comments posted directly on the pull request |
The cost of a review beyond the free tier
A developer runs /code-review with the medium effort level on a three line change touching an error message shown to the user. The review reports a single finding, about the wording of the message.
Write in one sentence what this situation establishes, and in one sentence what it does not establish.
What this establishes: This result establishes that the medium level of /code-review reported one finding on this precise three line change.
What this does not establish: It does not establish that a higher effort level, such as high or max, would not have reported any additional finding on the same change.
The three most common miscalibrations
- Too broad This result proves that the medium level is always enough for any code change, whatever its size.
- Too narrow This result says nothing since only one finding was reported on a change of just three lines.
- Beside the point This result shows that the error message concerned was already translated into several languages before the review.
- /code-review offers five local effort levels, low, medium, high, xhigh, and max, included in normal plan usage, typed directly after the command, with no flag in front of them.
- Low and medium only report the most certain findings, high, xhigh, and max widen the coverage at the cost of less certain findings.
- Ultra is not a sixth level of the same scale, it is a multi agent review in the cloud, billed separately from the local review.
- A third facility, reserved for Team and Enterprise plans, automatically comments on every pull request on GitHub, at a cost distinct from that of ultra.
- Asking for a git action in plain language actually triggers a command such as gh pr create, and forcing a push or deleting a branch remains subject to the same permission mode as any other command.
On a real change you have just made, run /code-review with an effort level proportionate to its risk, read each finding reported, then ask Claude Code in plain language to create the corresponding pull request.
Every datable claim in this lesson links here to the public text behind it. A source that does not open proves nothing.
- Claude Code, code review, /code-review effort levels and GitHub App service cost consultée le 2026-09-02
- Claude Code, ultrareview, free runs and usage credit cost consultée le 2026-09-02
- Claude Code, common workflows, creating a pull request in natural language consultée le 2026-09-02