Reliant Cloud
Hosting, releases and observability for apps built with agents.
Reliant Cloud is a hosting and delivery platform, like GitLab, Vercel and Supabase in one, that runs what you ship with Forge and monitors it.
It hosts your app and its database, builds each release once and promotes it through your environments, and watches what it runs, with the evidence your agent needs to fix what breaks.
Commands are shown as the Forge CLI spells them. Inside Reliant, run the same command as reliant forge ….
Part 1 of 6
Hosting
Reliant Cloud runs your services, workers, jobs, static sites and Postgres databases. Each environment is its own deployment with its own data, and one command ships it.
Services, workers and jobs hosted
ShippedServices, workers and jobs run hosted under a restricted profile: read-only root filesystem, no service-account token, quotas and egress containment.
forge env deploy prodStatus and detailsManaged Postgres
ShippedEach environment gets Postgres that the platform provisions and operates, on its own server, so no other tenant shares it and staging never holds a credential for prod’s data.
Status and detailsBackups and point-in-time restore
ShippedHosted databases archive WAL continuously and take a daily base backup. A point-in-time restore is a declaration: a new database booted from the source’s backups at the moment you name.
Status and detailsStatic hosting with a CDN
ShippedStatic sites are served from a platform-owned bucket behind a CDN, invalidated on each release, with per-environment runtime config that needs no rebuild.
Status and detailsCustom domains with automatic HTTPS
ShippedAdd a hostname and publish the DNS records Forge prints. The platform verifies it and issues, renews and serves its certificate.
forge domain add app.example.com --env prodStatus and detailsPer-PR preview environments
ShippedA pull request becomes its own environment, copied from a template env. Your CI creates, deploys and deletes it; an expired hosted preview is deleted by the platform.
forge env preview create --pr <n> --ttl <d>Status and detailsWrite-only hosted secrets
ShippedSecrets are write-only: set once, never read back, scoped to one organization and environment, in an OpenBao-backed store.
Status and detailsOne static site free
ShippedOne hosted static site is free, with no billing set up. Each environment’s site counts as one site.
Status and details
Limits
One hosted region
Hosted services and managed databases run in one region, GCP us-central1, with no region choice. Static sites are served through the CDN edge.
On the features pageHosted frontends are static exports
Reliant hosting serves a frontend only as a static export: no server-side rendering, server actions, middleware or request-time cookies. A Vite SPA or a statically exportable Next.js app fits as it is; one that needs a server runs as a container on a cluster you run.
On the features pageHosted backends do not autoscale
A hosted service or worker runs the replica count its env declares; it does not scale with traffic, and a managed database does not grow on its own. Raise the replicas or the shape in the env and deploy, or run the workload on a cluster you run with your own autoscaler.
On the features page
Part 2 of 6
Release management
A release is built once and promoted through your environments by digest. Each promotion is checked, recorded and judged, and an earlier release is one redeploy away.
Build once, promote by digest
ShippedA release is an immutable set of artifact digests, built once. Prod deploys exactly the bytes that passed staging.
forge env deploy prod --from stagingStatus and detailsProve each deploy is good
ShippedAfter a deploy, compares the window after the promotion with the equal window before it, per service, and exits pass, fail, could not tell or timed out.
forge obs check prod --waitStatus and detailsProd gated on staging's verdict
ShippedDeclare a
Status and detailsHealthGateand prod refuses a promotion until staging holds a passing verdict for that release, with an audited override for incidents.Required checks and your approvals
ShippedAn environment refuses a promotion that lacks its required checks, or that your own approval system did not approve, enforced where the environment’s ledger lives.
Status and detailsMigrations gate every rollout
ShippedMigrations run under a Postgres advisory lock before any new pod starts, so replicas never race to apply one and no pod runs on an unmigrated schema.
Status and detailsRollback floors per environment
ShippedRedeploy any earlier release, down to a floor. A release with a migration that breaks rollback raises the floor itself, so a redeploy never lands below a breaking schema change.
Status and detailsRelease timeline per environment
ShippedEvery promotion is an append-only entry: what was live at any moment, and what changed in it, from commits and pull requests to migrations, gates and health.
forge env releasesStatus and details
Part 3 of 6
Observability and monitoring
Logs, traces, metrics, errors and real-user sessions from every environment, tied to the release that produced them. Your agent reads the same evidence from the shell.
Query prod from the shell
ShippedLogs, traces and metrics answer from the shell, locally or for a hosted environment, as JSON with runnable next commands.
forge obs logsStatus and detailsTrace trees with uninstrumented time
ShippedA trace prints as a tree with errors and the critical path expanded, linked logs, and the time that went to uninstrumented code.
forge obs traceStatus and detailsError issues tied to releases
ShippedErrors group into issues with the culprit, frames, trace and suspect commits. A resolved issue reopens when a later release brings it back.
forge obs issues prodStatus and detailsProduction alerts out of the box
ShippedEvery environment starts with alert rules for error rate, latency, crash loops, failed deploys, new and regressed issues, uptime and certificates, delivered in the app, to Slack or to PagerDuty.
Status and detailsAlerts as code, run locally
ShippedAlert rules and routes are code, validated at render, recorded in the release, and evaluated locally through a real Alertmanager.
forge obs alerts checkStatus and detailsBad releases flagged automatically
ShippedEvery promotion is judged when its window closes, under the environment’s policy. A failing release raises
Status and detailsReleaseRegressedwith the findings and its commit range.Uptime and certificate probes
ShippedThe platform probes your public frontends and routes from outside, with nothing to declare, and alerts on failing checks and certificates about to expire.
Status and detailsReal-user sessions and replay
ShippedThe browser SDK reports through a same-origin relay that never forwards cookies or credentials, and one user’s visit prints as a timeline. Replay is opt-in and always masked.
forge obs sessionStatus and detailsDashboards derived from code
ShippedDashboards are generated from what you deploy, and an audit explains why an empty panel is empty.
forge obs dashboards auditStatus and details
Part 4 of 6
Testing
Tests run against real Postgres from the first commit, in unit, integration and end-to-end tiers, so a green run means the code works against the database it will meet in production.
Real Postgres in every test
ShippedDatabase tests run against a real embedded Postgres, each test on its own copy of a migrated template, with no Docker and no database mocks.
Status and detailsUnit, integration and e2e tiers
ShippedUnit tests use the real subject with generated mocks, integration tests a real migrated Postgres, and end-to-end tests a running dev stack.
Status and detailsTest reports for every run
ComingEvery run of your test lanes is recorded with its results, kept per commit and per release, and shown as a report.
forge test reportStatus and detailsCode coverage, merged across lanes
ComingCoverage from unit, integration and end-to-end runs merged into one report, per lane and combined, by package and by file.
Status and detailsFlaky-test detection
ComingEach test gets a flake score from its run history, so a test that passes and fails on the same code is named before it wastes a merge.
Status and details
Part 5 of 6
Performance
Production profiles show where time and memory go, and feed the compiler, so the code paths production runs hot get optimized.
Continuous CPU and heap profiling
ShippedEvery process samples a CPU profile every five minutes and a heap snapshot every two hours. It is on by default locally; one line in the env file turns it on for a hosted environment.
forge obs profileStatus and detailsProfile-guided builds from production
ShippedTurns the CPU profiles your environment collects into each Go binary’s
default.pgo, which you commit; every build then optimizes the call sites production runs hot.forge pgo pullStatus and detailsProfile-guided builds, refreshed weekly
ComingA scheduled workflow pulls the week’s CPU profiles and opens a pull request with every binary’s refreshed
Status and detailsdefault.pgo.
Limits
Hosted profiling is opt-in
Continuous CPU and heap profiling is on by default only in local dev. A hosted env collects profiles once its env file turns profiles on, which is one line.
On the features page
Part 6 of 6
Security
Dependencies are scanned, images are signed, tenants are isolated, and production data is read only through an audited path.
Dependency and image scanning
ShippedRuns govulncheck over every Go module and npm audit over every frontend, refuses a pass it did not verify, and takes reviewed, expiring exemptions instead of silent ignores.
forge ci vuln-scanStatus and detailsVulnerabilities by release and environment
BetaThe console’s Vulnerabilities page lists known vulnerabilities for each release and environment, and calls a release clean only when every artifact in it was scanned.
Status and detailsSigned images, SBOM and provenance
ShippedThe generated image workflow signs every image it pushes with a keyless signature bound to its digest, with an SBOM and build provenance attested.
Status and detailsAn SBOM recorded with every release
ComingEvery release cut records a CycloneDX SBOM per artifact, stored with the release, so its vulnerabilities can be checked again after it ships.
Status and detailsAudited read-only prod queries
ShippedRuns one statement in a read-only transaction, as a role that cannot read secret columns, and demands a reason; on a hosted environment it needs the organization’s
db:querygrant and records an audit row.forge db queryStatus and detailsDatabases fenced off by default
ShippedA hosted database accepts connections only from your own workloads and has no internet egress, by default.
Status and detailsTenant workloads sandboxed
ShippedHosted workloads run on an isolated runtime in per-organization, per-environment namespaces, and a cross-tenant test suite proves no query, secret or network path crosses tenants.
Status and details
Related
- Why ReliantOne platform that knows every seam, so your agents can too.
- All features: ReleaseBuild once, promote by digest, and gate each environment on the last one’s health.
- All features: RunHosting, data, secrets and telemetry for what you shipped.
- All features: FixAlerts and errors that reach an agent with the evidence to find the cause.
- Bring an existing appAny language, any repo: environments, releases and monitoring without a rewrite.
- ForgeOpen-source framework and CLI that puts guardrails on any app and deploys it with no account.
Ship your app on Reliant Cloud.
Deploy a static site or a service with no account, or let your agent deploy it. Claim it when you want to keep it.