Volcano Documentation

Volcano is AI infrastructure for agents and applications: functions, durable functions, sandboxes, databases, auth, realtime, storage, hosting, and locks.

Volcano is AI infrastructure for agents and applications. It gives you functions, durable functions, sandboxes, databases, authentication, realtime, file storage, frontend hosting, and distributed locks, all through one API.

Quick start

New to Volcano? Start here:

  1. Quickstart guide — Deploy your first function in 5 minutes
  2. Core concepts — Understand how Volcano works
  3. Installation — Set up the SDK and CLI

Ways to use Volcano

Everything Volcano does is one API. The dashboard, the CLI, the SDKs, and the MCP server are all ways of calling it — see Ways to use Volcano for which to reach for.

InterfaceReach for it whenDocumentation
DashboardExploring, one-off changes, reading logs and usageDashboard
CLIDeploying from a terminal or CI, migrations, local developmentCLI reference
SDKWriting application code — auth, queries, storage, realtime, durable functions, sandboxes, and distributed locksJavaScript, Python, Ruby
REST APIAutomating what the CLI does not coverAPI reference
MCP serverLetting an AI agent inspect or change one projectMCP server

Core features

Functions

Agents and APIs in Node.js, Python, or Ruby, invoked over HTTP and deployed to the project's regions, with cron schedules and the caller's identity on the request.

GuideDescription
OverviewWhat functions are and how they work
Creating functionsDeploy your first function
Invoking functionsCall functions from your app
Scheduled invocationsRun a function on a cron schedule
User contextAccess authenticated user data
Environment variablesConfigure secrets and settings
LogsView and filter function logs

Durable functions

JavaScript or Python workflows that checkpoint each step and resume for up to a year, with waits, retries, approvals, idempotent starts, and cron schedules.

GuideDescription
Durable functionsLong-running work that checkpoints and resumes
Durable approvalsPause a workflow until a person approves or denies it
Durable messagesSend messages to a running workflow and read what it publishes
Durable callbacksLet another system complete a waiting step with a token
Developing durable functions locallyRun one on your machine, then deploy it unchanged

Sandboxes

Isolated sessions where agents and applications run commands, keep files, and serve HTTP, from a published preset or a custom template.

GuideDescription
OverviewWhat Sandboxes do, who can use them, and how runtime is billed
Sandboxes guideSessions, files, HTTP access, user grants, and custom templates
Deploy a custom templateBuild your own image from a Dockerfile fragment
CLIRun commands and manage sessions from a terminal

Databases

PostgreSQL that auto-scales and branches in under a minute, with row-level security, a browser query builder or a direct connection, and backups with point-in-time restore on SUPERAGENT.

GuideDescription
OverviewDatabase features and access methods
Quick startSet up a database in 5 minutes
BranchingFork a copy of a database for every pull request, migration, or test run
Backups and restoreBack up a database and restore it to a backup or a point in time (SUPERAGENT)
Importing dataMove an existing PostgreSQL database in with pg_dump, and bulk-load rows with COPY
Query Builder APIQuery from the browser with the SDK
REST APIHTTP endpoints for database operations
Direct connectionConnect from a function with user impersonation
Row-level securitySecure data with policies
Auth helpersSQL functions for user context

Authentication

Email and password, Google, GitHub, Microsoft, and Apple, plus anonymous users that convert to a permanent account.

GuideDescription
OverviewAuthentication features and flow
QuickstartAdd auth to your app
ConceptsUsers, tokens, and sessions
OAuth providersGoogle, GitHub, Microsoft, Apple
Anonymous usersGuest access with upgrade path
Password resetForgot password flow
ConfigurationCustomize auth behavior

Storage

Buckets that are private by default, with access policies, per-file public links, and resumable uploads.

GuideDescription
OverviewStorage features and quick start
JavaScript SDKComplete SDK reference
PoliciesRLS-style access control
BucketsCreating and managing buckets

Distributed locks

Renewable leases so one holder runs a task, with a fencing token the protected resource can check.

GuideDescription
JavaScript SDKCoordinate backend workers with renewable project leases

Realtime

Postgres changes, broadcast, and presence over WebSockets.

GuideDescription
OverviewRealtime features and the connection model
Postgres changesSubscribe to database changes
BroadcastSend messages between clients
PresenceTrack who is online
JavaScript SDKClient SDK reference
SecurityChannel authorization and limits

Frontends

Next.js, static or server-rendered, served from the edge. A failed build never replaces the live version. Custom domains are on every plan.

GuideDescription
OverviewHow frontends build, deploy, and serve
Deploy a frontendDeploy a Next.js site with the CLI
Frontend previewsDeploy a build to its own URL, then promote or roll back

Security

GuideDescription
Anon keysPublic keys for frontend use
Service keysSecret keys with admin access
Project access tokensProject-scoped API credentials for CI, scripts, and agents
Token typesUnderstanding different token types
Security checklistProduction security guide

Projects

GuideDescription
Deploy from GitHubConnect a repo and deploy on every push to the production branch
Export your source to GitHubInitialize an empty repository with a project's stored source and deploy it from Git
Configuration manifestDeclarative volcano-config.yaml reference (config deploy/pull)
Variable EnvironmentsCreate and manage Project variable Environments
Project healthCheck project health and review operation failures and blocked storage use
NotificationsGet an email or a signed webhook when a scheduled function keeps failing

API reference

Complete REST API documentation for all Volcano endpoints.

ReferenceDescription
OverviewAPI basics and authentication
Using the APICreate an access token, deploy, read logs, handle errors
OpenAPI specificationFetch the machine-readable contract and generate a client
AuthenticationAuth headers and token types
Auth endpointsSignup, signin, and user management
ProjectsProject management endpoints
FunctionsFunction deployment and invocation
DatabasesDatabase provisioning endpoints
StorageStorage bucket and object endpoints
Project locksBackend lease endpoints
NotificationsAlert settings, webhook secret rotation, and test sends
ErrorsError codes and handling

Guides

Step-by-step guides for common tasks.

GuideDescription
Deploy to productionShip your project to the cloud
Plans and limitsHOBBY vs SUPERAGENT limits by resource
Database migrationsApply schema changes once each, rehearsed and backed up, or with the CLI
Deployment performanceDeployment latency objectives, phases, and telemetry
SandboxesRun commands and HTTP services in isolated sessions with plan-based runtime pricing
IssuesSee runtime errors grouped into issues, and resolve or ignore them
LavaFlowHave AI diagnose your issues, open pull requests that fix them, and merge them

Examples

Working code examples in the examples/ directory:

  • nodejs-hello/ — Simple Node.js function
  • python-data/ — Python data processing
  • frontend-auth-nextjs/ — Next.js frontend with authentication
  • function-sdk-example/ — Using the SDK inside a function
  • realtime-chat/ — Real-time chat with presence tracking

Getting help

  • API reference — REST API endpoints, authentication, and errors

On this page