The check reported branch protection fully green while a repository ruleset on the same branch required two status checks that no workflow produces — 'Build, Lint & Test' and 'All checks passed', names left behind when the jobs were renamed. Nothing could merge without --admin, and the report said everything was fine. That is the worst shape a check can take: confident and blind. A branch can be governed by classic protection and a ruleset at once, independently, and the stricter wins. Reading one and calling it the answer is how a broken gate stays invisible for months while the team learns to route around it. --check now reads active rulesets too, and compares each required check against what recent pull requests actually report. A rule waiting on a check that never arrives does not slow a branch — it closes it. Sampling uses recent PR heads rather than main's head, because the aggregate gate only runs on pull_request; sampling main would report it as never seen and manufacture the opposite false alarm.
7.0 KiB
ops
Operational configuration that lives outside the application code.
apply-repo-settings.sh
The repository's GitHub protections, as code.
Branch protection and the production environment are configured through the
GitHub API. Nothing in the repository reflects them, so by default they are
invisible here, unversioned, and gone without a trace if someone removes them —
which undercuts the point, since both exist precisely so that nothing reaches
main or production unreviewed.
./ops/apply-repo-settings.sh # apply the desired state
./ops/apply-repo-settings.sh --check # report drift, change nothing
Idempotent: each call PUTs the full desired state, so repeated runs converge.
Requires gh authenticated with admin rights on the repo.
What it enforces
Branch protection on main
| setting | value | why |
|---|---|---|
| required check | CI Gate ✅ |
aggregate job — fails if web, backend or security-audit did not succeed |
| strict | true | branch must be up to date with main before merging |
| approvals | 0 | the PR is the gate, not a second reader — see below |
| dismiss stale reviews | true | an approval does not carry over to new commits |
| force pushes | blocked | history cannot be rewritten |
| deletions | blocked | main cannot be deleted |
| conversation resolution | required | review comments cannot be merged past |
Only CI Gate ✅ is listed as a required check, deliberately. It runs with
if: always() and inspects the results of the jobs it needs, so it already
covers them. Listing each job here as well would mean editing this file whenever
one is renamed — and a rename would then quietly drop a requirement instead of
failing loudly.
Why zero required approvals
Requiring an approval on a four-person team means every change waits on somebody
else being awake. That cost buys less than it looks like: the thing worth
protecting here is not code review — it is that nothing reaches production
unnoticed, and that funds cannot move on one person's say-so. Those live
elsewhere. Production deploys wait on the production environment, and
fund-moving operations are being put behind on-chain multisig, where a stolen
account or a compromised host cannot forge a second signature the way a
collected token can forge a second approval.
What the branch rule still buys, at zero waiting: nothing lands on main
without a PR, and nothing merges with red CI. That is the part a four-person
team actually gets value from.
enforce_admins stays false on purpose: an admin needs a way through during a
real incident. It is an escape hatch, not the normal path — rare, visible, and
recorded in the branch's protection history.
production environment
Deployment approval is an environment protection rule, not something the workflow implements. GitHub holds the run until a reviewer approves it:
- no runner is occupied while waiting
- the window is 30 days, so a missed notification costs nothing
- who approved which SHA is recorded in the deployment history
- approval state survives a runner restart
Exactly one job in quality-gate.yml opts in with environment: production:
approve-deploy. The four deploy jobs are held by depending on it.
notify-deploy-pending only sends a Telegram message pointing at the run; it
cannot block anything.
This replaced a job that polled /tmp/pexsec-gates for 30 minutes while holding
a runner slot. On 2026-07-30 that window expired unseen and cancelled a deploy —
a merged migration never reached the database and the only trace was a failed
job.
Why one gate job and not environment: on each deploy
The first version put environment: production on all four deploy jobs, which
reads as the more direct design — GitHub itself holds each one. It does not
behave that way.
GitHub opens an approval request for the jobs that are eligible at that moment,
and these four never are. deploy-supabase waits only on the notification;
the rest wait on image builds that take minutes. So a run asks twice. On
2026-07-31 the first approval shipped the Supabase functions and migrations, and
the run then went back to waiting — unnoticed, because the notification had
already been sent and answered. The schema had moved and the app serving it had
not. Nobody is told about the second round; you have to be looking at the run.
One gate job removes the split by construction: everything that deploys hangs off a single approval, so there is only ever one to give.
The cost is that approval is now enforced by a dependency edge rather than by
GitHub. ops/check-deploy-gate.py restores the guarantee — it fails CI if a
deploy-* job does not depend on approve-deploy, if a second job declares the
environment (which brings the split back), or if a job uses always() without
asserting the gate succeeded, which silently un-holds it. The Deploy gate wiring job runs it on every PR.
python3 ops/check-deploy-gate.py
Known: the version bump lands before approval
bump-version commits web/package.json to main and pushes it before the gate
— so rejecting a deploy leaves a bumped version for a release that never shipped.
It is a number in a file, [skip ci], and touches no server. Left as is rather
than changed quietly; moving it behind approve-deploy is a one-line change if
that is preferred.
Stopping a run that is waiting for approval
gh run cancel does not work on a run held at an environment gate — it reports
success and the run stays waiting, still holding the queue behind it. The
pending deployment has to be rejected:
gh api repos/:owner/:repo/actions/runs/<RUN_ID>/pending_deployments \
-X POST -f state=rejected -f comment='superseded' \
-F 'environment_ids[]=<ENV_ID>'
Get <ENV_ID> from gh api repos/:owner/:repo/actions/runs/<RUN_ID>/pending_deployments.
This came up on 2026-07-31: a stale run sat in waiting and blocked the deploy
of the run behind it until it was rejected.
Two layers, one branch
A branch can be governed by classic protection and a repository ruleset at
the same time, independently, and the stricter of the two wins. That is how
main here spent months unmergeable without anyone knowing why: classic
protection asked for CI Gate ✅, while a ruleset on the same branch asked for
Build, Lint & Test and All checks passed — two names no workflow has
produced since the jobs were renamed. Every merge went through --admin, and
the habit read as "review is slow" rather than "the gate is broken".
--check now reports both, and flags a required check that nothing reports.
A rule demanding a check that never arrives does not slow a branch down; it
closes it, silently, and the drift report will say everything is fine.
Reviewer
REVIEWER_ID is a numeric user id rather than a login, because the API takes ids
and a login rename would silently break the rule instead of erroring.
Override any of it via environment variables:
REPO=owner/repo BRANCH=main REVIEWER_ID=123456 ./ops/apply-repo-settings.sh