Name it. Choose a plan. Create it.
The first screen asks for the few decisions that change the service. Database credentials and the endpoint are generated after provisioning.
- EnginePostgreSQL 17
- Plan priceshown before create
- ConnectionTLS required
Automatic provisioning via MCP & REST, no ticket queue. Hard-capped fixed pricing from $1/mo protects you from runaway autonomous loop bills. PostgreSQL 17 with TLS and PgBouncer pooling.
npx -y @stashidb/mcp-server{
"tool": "create_checkpoint",
"params": {
"label": "before orders.status migration"
}
}postgresql://agent_scoped_role:••••••••@db.stashi.dev:6432/orders_api?sslmode=requireWhen AI agents write schema, execute autonomous migrations, or store multi-agent memory, serverless compute meters can bankrupt you overnight. Stashi combines fast, automatic provisioning with hard-capped flat pricing and one-command checkpoint rollbacks.
Autonomous reasoning loops running 24/7 will never spike your card. Hard billing caps at $1, $3, or $5/mo guarantee 0% runaway cloud bills.
Every plan gets its own role and schema, provisioned the moment you ask — no approval queue, no ticket, no human waiting on the other end for agent test runs or new tenant databases.
If an autonomous agent hallucinates a faulty migration or drops a critical relation, restore the last checkpoint with one command — no manual pg_restore, no DBA on call.
Agent: 'Inspect existing tables and row counts before writing the migration.'
list_tables()SELECT c.relname AS table_name, c.reltuples::bigint AS estimated_rows,
pg_size_pretty(pg_total_relation_size(c.oid)) AS total_size
FROM pg_class c JOIN pg_namespace n ON n.oid = c.relnamespace
WHERE c.relkind = 'r' AND n.nspname NOT IN ('pg_catalog', 'information_schema')
ORDER BY c.relname;{
"mcpServers": {
"stashi": {
"command": "npx",
"args": ["-y", "@stashidb/mcp-server"],
"env": {
"STASHI_API_KEY": "st_live_9f8a2bc4e107",
"STASHI_DATABASE_ID": "DB_xxx"
}
}
}
}The dashboard is built around the work developers and agents actually do: provision in milliseconds, copy TLS credentials, watch node saturation, restore instantly.
The first screen asks for the few decisions that change the service. Database credentials and the endpoint are generated after provisioning.
postgresql://inventory_owner:••••••••@db.stashi.dev:6432/inventory?sslmode=requireCopy the URL into Prisma, Django, Rails, LangChain, Cursor or psql. Credentials can be rotated from the same database view.
Stashi tracks node pressure and database limits in the operator layer. Customer plans stay simple while the service watches CPU, memory, disk and connection pressure underneath.
Start with the workload you have. Storage, connection limits and backup retention are published with each plan.
Compare plansAgent sandboxes, experiments, and scratchpads.
Small apps and multi-agent memory.
High-concurrency swarms and active apps.
Continuous reasoning & reserved capacity.