[7ttp]

TL;DR

makes config push apply [auth.sms] enable_confirmations the way start and email confirmations do

whats hurting the user?

config push sent enable_confirmations = true as sms_autoconfirm = true, so phone signups and
updateUser({ phone }) skipped the phone code

config diff and pull read it back the same wrong way, so diff stayed clean

now fixed by?

push, diff and pull now send and read sms_autoconfirm as the opposite of enable_confirmations,
like email already does

visual rep

%%{init: {"flowchart": {"wrappingWidth": 400, "rankSpacing": 60, "nodeSpacing": 30}}}%%
flowchart LR
    cfg["enable_confirmations = true"]
    cfg -- before --> oldBit["sms_autoconfirm = true"] --> oldOut["phone code skipped"]
    cfg -- now --> newBit["sms_autoconfirm = false"] --> newOut["phone code required"]

Loading

ref:

• closes: config push sends [auth.sms] enable_confirmations un-negated as sms_autoconfirm, so hosted projects get the opposite of local #6997
• closes: CLI-2624

BREAKING CHANGE: [auth.sms] enable_confirmations = false, the init default, now turns phone confirmation off on the next config push, set it to true to keep the phone code and run config diff before pushing

[github-actions]

🤖 AI Review

The SMS inversion is correct and consistent across push, pull, and local start. Claude's minor migration advisory is confirmed only for explicitly declared false values; omitted keys are never pushed. No additional findings were identified. Tests were not run because dependencies are unavailable.

Findings

Severity	Location	Category	Sources	Claim
🟡 MINOR	packages/config/src/project-config/registry-auth.ts:785	behavior-change	claude	For existing projects explicitly declaring auth.sms.enable_confirmations = false while hosted sms_autoconfirm is false, the next approved config push now sets sms_autoconfirm = true, disabling phone-number confirmation. Although this correctly implements the setting, users relying on the previous mapping need a migration note explaining the security-relevant behavior change.

Stats

Claude findings: 1 · Codex findings: 0 · Confirmed: 1 · Refuted: 0 · Uncertain: 0

---

Models: claude-opus-5-5 + gpt-6.1-sol · Trigger: auto · Workflow run

This review runs once per PR. A maintainer can request another with a /ai-review comment.

[Coly010]

Thanks for picking this up! I ran it against a staging project and the true case now behaves the way you'd expect

The one thing I think needs to change is where it lands. supabase init writes [auth.sms] enable_confirmations = false, so for a project still on that default whose hosted sms_autoconfirm is false, config diff goes from clean on develop to:

auth.sms.enable_confirmations [update]
  local:  false
  remote: true

and the next config push --yes (pretty common in CI) turns phone confirmations off. That's the correct reading of the config, but it's the same config.toml producing a different hosted result, so I think it falls under the breaking-change rule in AGENTS.md:

1. retarget to next and retitle as fix(config)!: ...
2. add a BREAKING CHANGE: footer at the very end of the PR body with the migration note, your existing NOTE text works fine for it

The NOTE in the description is good, but on develop it never makes it into the release notes, so the folk most affected wouldn't see it.

If you think the security side means it should ship on develop sooner, I'm open to that conversation, I'd just rather we didn't flip it quietly for everyone on the init default. Otherwise it looks good to me!