Scheduled maintenanceinstanode is temporarily unavailable while we perform maintenance. Your data is safe and we’ll be back shortly. Thanks for your patience.
Skip to main content
Heroku·but for AI agents

Real infrastructure your agent can claim with one curl.

Postgres, Redis, MongoDB, vectors, queues, storage, webhooks, and deployments. Provisioned in <2 seconds. Self-serve from the first call to a paid plan — no signup, no Docker, no sales call.

Paste this into Claude Code, Cursor, or any AI coding agent

Build me a tiny expense tracker — Postgres backing store, a FastAPI app on top, and deploy it to a public URL I can hit from my phone. Use instanode.dev — no signup, provision whatever you need.

The whole stack, not the pieces.

Free Postgres alone is commoditized. Free Postgres plus Redis, Mongo, queue, storage, webhook, and a real deployment — claimed in one in-flow token — is the wedge.

Postgres
POST /db/new
live in 1.4s
Redis
POST /cache/new
live in 0.9s
MongoDB
POST /nosql/new
live in 1.2s
Vector (pgvector)
POST /vector/new
live in 1.4s
Queue (NATS)
POST /queue/new
live in 0.7s
Storage (S3-compatible)
POST /storage/new
live in 0.8s
Webhook
POST /webhook/new
live in 0.3s
Deploy
POST /deploy/new
live in ~60s

From prompt to production, in one chat.

Anonymous tier with a 24 h TTL. The agent surfaces a claim link when the quota tightens. One click → magic link → real dashboard.

STEP 01 · CALL

Agent calls.

POST /db/new from inside Claude Code, Cursor, your custom MCP tool, or a shell script. No keys. No setup.

STEP 02 · PROVISION

Real infra spins up.

Managed Postgres, Redis, or Mongo — encryption at rest for vault secrets and stored credentials, a connection string in 1.4 s. 24 h TTL on the free tier.

STEP 03 · CLAIM

Claim when ready.

When the agent surfaces the claim link in its response, one click → permanent ownership. Resources inherit your team’s tier.

From throwaway test DBs to agent swarms.

Coding agents that need memory, hackathon teams that need a backend in three curls, multi-agent frameworks that need a message bus, indie founders shipping on Sunday night. Six exemplar shapes below; the full catalogue covers 100+.

Build note · 4 min readSixty seconds, one prompt, one working appDevika, a non-developer, asked her agent to build an expense tracker. Postgres + FastAPI + a real HTTPS URL. Three curls and the build took 90 seconds — including the deploy. Read the walkthrough →

Self-serve sign-up at every tier. No sales call.

Anonymous is the funnel. Hobby pays for the side project. Pro lets one product run in production + staging + development. Team is for the company that ships every day — talk to us about your needs.

Anonymous

For the first call. Before there is a team to bill.

free· 24 h TTL
  • 10 MB Postgres · 2 conn · 24h TTL
  • 5 MB Redis · 5 MB Mongo · 24h TTL
  • 100 stored webhooks · 0 deployments
  • no vault — claim resources first
See the curl ↗
Hobby

For the side project at 2 a.m. Same agent flow, just doesn’t expire.

$9/ mo
  • 1 GB Postgres · 8 conn
  • 50 MB Redis · 100 MB Mongo · 5 conn
  • 1 deployment · *.deployment.instanode.dev
  • 20 vault entries · production env
Start hobby →
Team

For the engineering org. Dedicated infra, RBAC + audit, with SSO & SLA on the way.

$199/ mo
  • 50 GB Postgres · 1.5 GB Redis · 40 GB MongoDB
  • 100 deployments · 50 custom domains
  • 90-day backups · self-serve restore (Redis backup-only) · RBAC + audit
  • Need more? Enterprise — contact sales
Contact sales →

Designed for the place your user actually is.

The first surface is the agent itself — a chat card, a JSON shape, a curl line. Drop us into Claude Code, Cursor, or any MCP-compatible client.

Claude Codecli
// inside Claude Code chat /instanode spin up a postgres for the flashcards app

Surfaces the claim card directly in chat and writes a .env for you.

MCP serverjson
// claude_desktop_config.json { "mcpServers": { "instanode": { "command": "npx", "args": ["-y", "instanode-mcp"] } } }

Eight provisioning tools: postgres, vector, redis, mongo, queue, storage, webhook, deploy — plus stack & deployment management (create_stack, list_stacks, update_stack_env, list_deployments, get_deployment, redeploy, delete_deployment, …).

curl / shellsh
# pipe straight into psql: psql $(curl -s api.instanode.dev/db/new -d '{"name":"prod-db"}' | jq -r .connection_url)

One line. No SDK, no client, no setup. Cursor, Aider, and any shell-based agent work the same way.