$ the-wire · showcase
Supabase C# ships the full passkey and MFA recovery code surface, docs get last-updated dates
By RepoJournal · Filed · About Supabase · Composed from the cited sources · methodology
![supabase-vercel-tedd[bot]](https://avatars.githubusercontent.com/u/54469796?v=4)
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?v=4)
10 people shipped this
Jordan McQueen
@jmqd
2 cited
supabase-vercel-tedd[bot]
@supabase-vercel-tedd[bot]
1 cited
kodster28
@kodster28
1 cited
supabase-releaser[bot]
@supabase-releaser[bot]
1 cited
Miranda Limonczenko
@czenko
Wendie Cheung
@wendeh
github-actions[bot]
@github-actions[bot]
Tr00d
@Tr00d
supabase-evals-releaser
@supabase-evals-releaser
supabase-evals-releaser[bot]
@supabase-evals-releaser[bot]
supabase-csharp v8.8.0 adds MFA recovery code generation, verification, and regeneration, while passkey list and delete landed as separate PRs, and the docs site now renders a last-updated date on every page.
v8.8.0 supabase/supabase-csharp
The 8.8.0 release lands MFA recovery codes in the C# client: GenerateRecoveryCodesAsync and a verify path that reaches aal2. RegenerateRecoveryCodesAsync replaces a signed-in user's codes at aal2 and invalidates every previous code at once.
feat(docs) - add last updated date supabase/supabase
Every docs page now renders a last-modified date pulled from a Supabase table, read through a prebuilt date map and covered by tests on the fetching script. The motivation in the PR is that "dates are a really important signal for LLMs."
chore(deps): bump follow-redirects to 1.16.2 (#52612) supabase/supabase
follow-redirects moves from 1.16.1 to 1.16.2 in pnpm-lock.yaml to clear the high-severity GHSA-mrwp-f3q8-3g9p advisory (CVE-2026-116689), which affects 1.16.0 through 1.16.1. The package is transitive, so the lockfile bump is the whole fix.
feat(xtask): Add CI metrics collector supabase/etl
cargo x ci-metrics collects GitHub Actions workflow runs, jobs, benchmark manifests, and code-scanning results into Postgres across two schemas, ci for raw data and ci_report for the views Grafana reads, plus per-test results from nextest JUnit for flaky-test and latency tracking.
test(ducklake): check FULL deletes match values supabase/etl
DuckLake finds rows to delete by comparing SQL text and never checks how many rows a FULL delete hit, so a stored value that disagrees with its text leaves the old row behind with no error. The new test pushes 42 tricky values through every write path and asserts each row is gone.
Action items
- → Bump follow-redirects to 1.16.2 wherever your lockfile resolves 1.16.0 or 1.16.1 (CVE-2026-116689) supabase/supabase [plan]
