GitHub push events not triggering deployments, need assistance - Vercel Staff? - Help

GitHub push events not triggering deployments --- exhausted all troubleshooting, need Vercel staff assistance

Framework: Next.js Git Provider: GitHub (MercuryThrives/elt-clarity-quiz)

What happened: Automatic deployments were working normally on April 22, 2026. On April 23, a GitHub token invalidation warning appeared in GitHub Desktop. After re-authenticating, pushes to main stopped triggering Vercel deployments entirely.

Symptoms:

Pushes to main produce no new deployments in the Vercel dashboard --- not even failed ones
GitHub commits show ✕ 0/1 on all commits since April 23
No Vercel Bot comments appearing on any GitHub commits
No webhook exists in GitHub repo Settings -> Webhooks, and none is created when reconnecting

Troubleshooting completed (in order):

Re-authenticated GitHub in GitHub Desktop after token invalidation warning
Disconnected and reconnected GitHub repo in Vercel Settings -> Git multiple times
Verified Vercel GitHub App at github.com/settings/installations --- set to "All repositories" with full read/write permissions
Confirmed no webhook exists in GitHub repo Settings -> Webhooks
Verified vercel.json does not contain git.deploymentEnabled: false
Fixed commit author email mismatch --- git config email now matches GitHub primary email
Matched Vercel account email to GitHub primary email
Reauthenticated GitHub connection at vercel.com/account/authentication --- shows MercuryThrives connected, last used today
Multiple test pushes after each fix --- no deployments triggered

Current state:

vercel.com/account/authentication shows GitHub/MercuryThrives connected
Project Settings -> Git shows repo connected
Git commit author email, GitHub primary email, and Vercel account email all match
No webhook in GitHub repo
Zero response from Vercel on any push

Request: This appears to require backend intervention to reset the webhook/integration state. Could a Vercel staff member investigate why the GitHub App is not creating a webhook for this project despite a valid connection showing on both sides?

Full commit SHA of most recent test push available on request. 7a300f1f98d725e641b3ef4c5cf769f2e3ff0570

Hitting the same family of issue here, with one extra data point that may help Vercel staff narrow it down:

My setup: monorepo with 3 Vercel projects (gg-public-site, gg-dashboard, gg-super-admin) all pointed at the same GitHub repo, same main branch, same GitHub App installation. Hobby plan.

Symptom: Pushes to main auto-deploy gg-dashboard and gg-super-admin normally. gg-public-site shows nothing — no queued, failed, or skipped build, just zero rows under Deployments. Started silently around 2026-04-25 (no token invalidation, no settings change).

The interesting part: since the GitHub App is the same and the other two projects on the same install are receiving events fine, this isn't an app-wide token/connection issue (which is what most of the related solved threads point at — disconnect/reconnect, GitHub App permissions, branch protection). It looks like a per-project webhook subscription that quietly got dropped server-side, not a repo-wide or org-wide one.

Already tried (none woke it up):

Empty commits, real file changes inside apps/public-site/, top-level changes
Disconnect + reconnect Git in Project Settings → Git
Verified GitHub App has access to the repo (other 2 projects on the same install work fine, so this is a sanity-check only)
Confirmed Ignored Build Step is empty and Skip Unaffected Projects matches the working projects
npx vercel --prod from the linked repo works fine, so the project + build are healthy

Workaround: CLI deploy via npx vercel --prod from a script. Not a fix, just unblocks shipping.

This is the third or fourth thread in the last few months reporting the same shape (per-project deploy hook silently going dead), so it would be great if a staff member could investigate whether there's a class of webhook subscriptions getting dropped without any user-visible signal. Happy to share project ID and a few SHAs that should have triggered, via DM.