Orbit
Orbit Project Settings
Every knob for an Orbit project lives on the Settings tab, grouped so you can find things: general information, build and deploy, security, notifications, integrations, usage limits, deploy automation, networking, observability, sharing, and a danger zone at the bottom.
Where Settings Lives
Open Orbit, click the project, and choose Settings in the project tab strip. The page describes itself as configuring your project name, build commands and Git integration, which undersells it considerably.
This guide is a map. Where a subject has its own article, it is linked rather than repeated.
General
Project info shows the URL slug, the creation date, the connected repository and the production URL, and lets you change the project Name and Description.
Git holds the Production branch. Pushes to that branch trigger production deploys.
Status badge gives you a live deploy-status SVG with copy buttons for the URL, markdown and HTML, ready to paste into your README. See Viewing Your Project README In Orbit.
Public status page links through to the status page configuration. See Orbit Status Page.
Build and Deploy
Build settings are the five fields Orbit auto-detects and you can override: install command, build command, output directory, root directory and Node.js version. Leave a field blank to keep the detected value. Full detail is in Configuring Your Build Command and Output Directory.
Runtime carries four switches worth understanding:
- Server mode keeps the build machine alive running your start command after each deploy. Required for anything that renders on the server rather than exporting static files.
- Branch previews create an environment automatically for every branch push that is not production or staging, each on its own URL under
branch-*.kaps.run. See Branch Previews. - Auto-rollback on failure restores the last healthy production deployment if a deploy fails.
- Scheduled rebuild rebuilds production on an interval, from hourly to weekly, which suits content-managed sites that need a refresh without a git push.
Build auto-retry re-queues builds that failed on infrastructure errors such as a network problem or a timeout, up to three times. It does not retry code errors, so it cannot mask a broken build.
Ignored paths and Branch ignore patterns stop builds you do not need: glob patterns for files whose changes should not deploy, and for branches such as dependabot/* that should never trigger one.
Git tag deploys sets the tag pattern that turns a tag push into a production deploy. See Orbit Releases.
Deploy Automation and Protection
Deploy protection has two gates: Require approval for production, which pauses push-triggered production deploys until someone approves, and Require staging success before production, which holds production until staging has deployed the same commit successfully.
CI required checks gates deploys on your own CI. On GitHub you list job names, all of which must pass; on GitLab any non-empty value waits for the whole pipeline. A CI failure cancels the deploy.
Deploy freeze schedule blocks push-triggered deploys outside approved windows, either weekends or a UTC hour range. Manual deploys and deploy hooks are unaffected.
Branch protection adds glob-pattern rules that block deploys on matching branches until required external checks pass and, optionally, a human approves.
Auto-promote staging promotes staging to production automatically once it has run for a configured number of hours with no health failures and smoke tests passing.
All of these are summarised in one view on the Orbit Deployment Pipeline tab.
Health and Quality Gates
Health check fetches a path you choose after every production deploy. A non-2xx response within 15 seconds restores the previous healthy deployment.
Smoke tests run GET requests against up to ten comma-separated paths after each successful deploy and record pass or fail. Combined with auto-rollback, a failing smoke test rolls the deploy back.
Performance budgets warn or fail builds that exceed a threshold: total artifact size in MB, individual file size in kB matched by a glob, or build time in seconds. Each budget is set to either warn or fail the build.
Start every budget as a warning. Run it for a fortnight, see how often it trips, then promote the ones that were right to fail the build. A budget introduced as a hard failure on day one usually gets deleted the first time it blocks a deploy at a bad moment.
Staging
Staging gets its own set of cards: the staging branch, access protection with a password, an IP allowlist accepting IPv4 addresses and CIDR ranges one per line, auto-rollback on failure, whether staging inherits production environment variables at lower priority, and build overrides for install, build, output and root directory.
Password and IP allowlist are the two to set on any staging environment that is not meant to be public. Leave both blank and the staging URL is reachable by anyone who has it.
Networking
Custom domains adds and verifies domains for the production environment, including certificate provisioning. See Adding a Custom Domain to Orbit.
Redirects and rewrites are path-based rules applied before serving, tested in order with first match winning. A redirect sends the browser to a new URL as a 301 or 302; a rewrite serves a different path silently with no URL change. Patterns support exact matches, wildcards, named parameters and splats. See Redirects and Rewrites in Orbit.
Response headers apply custom HTTP headers to matching paths, with presets for HSTS, CSP, no-embed, no-sniff, referrer policy and CORS. All matching rules are applied, and later entries overwrite earlier ones for the same key.
Skew protection keeps old build artifacts in storage for a retention window after a new deployment goes live, so a visitor who loaded the previous version can still fetch its assets instead of getting a 404.
The config store holds key/value settings your deployment reads at runtime, for feature flags and settings that change without a redeploy. It is managed under Storage, not here, and it is free: there is no read charge and no write charge. Values are stored in plain text, so never put a secret in one. Up to 100 entries per project, keys up to 200 characters and values up to 4096. A nightly snapshot of every entry is taken and thirty nights are kept on every plan, so an entry you delete can be brought back: see Snapshots and restore.
Usage and Limits
Build usage shows the builds this project has run.
Spending cap sets the maximum monthly overage spend, covering build-minute and bandwidth overages beyond your plan allocation. The minimum is US$6.32 and the maximum US$6,315.79. Alerts fire at 50% and 80%, and at 100% serving pauses until you raise the cap. See Orbit Spending Cap.
Build spending cap is the separate, blunter limit: a monthly ceiling on total build minutes, after which new builds stop queuing until next month or you raise it.
Artifact retention sets how many successful build artifacts to keep per environment. Older ones are deleted daily, and the currently live deployment is always kept regardless of the limit.
Preview expiry pauses preview environments after 7, 14, 30 or 60 days, or never. Paused previews have their storage reclaimed within 24 hours.
Artifact size alert emails you when a build artifact exceeds a size you set.
Notifications, Integrations and Observability
Deploy email notifications has three settings: all deploys, failures only, or off. A notification webhook URL can be set alongside it, posting on every success or failure.
Notification channels is the richer version, with per-event selection, delivery history and a test button. Full webhook control is on its own tab: see Orbit Webhooks.
Deploy hooks creates secret URLs that trigger a deploy when something POSTs to them. See Triggering Deployments Via Deploy Hooks.
Log drains ship build logs to your observability platform after each deployment, with built-in support for Datadog, Logtail, Axiom and New Relic, or a custom webhook, optionally filtered to one environment.
Web Analytics gives you the Core Web Vitals collection snippet. See Orbit Web Vitals.
Cron triggers schedules HTTP requests to a chosen environment, up to ten per project. The larger, production-only version is on its own tab: see Orbit Cron Jobs.
Sharing and Team
Team manages collaborators with three roles: admin, developer and viewer.
Share links generate temporary passwordless links so an external reviewer can view an environment without a password or an account. Each link can expire in 1 hour, 24 hours or 7 days, or never, and shows its visit count. Revoke one at any time.
Deployment group joins this project to a group so related projects deploy together, optionally filtered by branch.
Clone environment and Clone project duplicate configuration. Cloning a project copies build settings, environment configuration and non-secret environment variables; secrets are not copied.
Share links are passwordless by design. Anyone with the URL sees the environment. Set the shortest expiry that suits the review, and revoke it when the review is done rather than leaving a permanent public door into staging.
Danger Zone
Clear build cache deletes the cached dependencies for every environment, so the next deployment runs a full install from scratch. It cannot be undone, though the cache rebuilds itself on the next build.
Transfer project moves the project to another Kapsule account. See Transferring an Orbit Project.
Archive project stops serving traffic immediately. It is restorable within 30 days from the Orbit list, after which it is permanently deleted.
Archiving takes the site down the moment you confirm it. There is no grace period on the serving side, only on the deletion side. If you want a project to stop deploying but keep serving, lock deploys instead: see Deploying Your Project.
Where To Go Next
- Orbit Deployment Pipeline to see every setting's effect in one view.
- Shared Environment Variable Groups for values used across projects.
- Orbit Plan Limits for what your plan allows.