Drops

The language your AI agent uses to talk to InfiniteOcean. Plain sentences that save data, build websites, send emails, and run code. You don't need to learn it — your AI already knows it. But here's what it looks like.

Connect your AI Full Spec Docs

What It Looks Like

-- Save a user
save users @alice name: "Alice" role: "admin"

-- Get them back
users @alice

-- Find all admins
find users where role = "admin" sort name limit 10

-- Pick specific fields
find users where role = "admin" then pick name

-- Run some code
save fn/greet @main "def main(ocean, request):
  return {'hello': request.get('name', 'World')}"
save fn/greet @_config execute: "public"

-- Schedule it to run every morning
every day at 9am daily-greet: run fn/greet

-- Share access
share users with @collaborator-key

Try It Live

This is real. Every command runs live. No signup needed.

Drops Commands

How It Works

Send your commands as text. Each line runs in order. Results come back instantly.

# curl
curl -X POST https://api.infiniteocean.io/exec \
  -H "Content-Type: application/json" \
  -H "X-Ocean-Key: YOUR_KEY" \
  -d '{"query": "save users @alice name: \"Alice\"\nusers @alice"}'

# Response
[
  { "ok": true, "drop_id": "uuid", "key": "alice" },
  { "ok": true, "data": { "name": "Alice" } }
]

Multiple commands run top to bottom. If one fails, the rest still continue.

Every Command

Store & Retrieve

Save, get, update, and remove records.

save users @alice name: "Alice" users @alice update users @alice age: 30 remove users @alice users

Find & Search

Filter, sort, full-text search, semantic search, aggregate.

find users where age > 10 sort name find users sort created_at desc search users "alice" mode text find orders group category sum amount count count users

Run Code

Deploy Python functions, run them, fetch URLs.

save fn/x @main "def main(o,r): ..." run fn/x input: "data" async status "job-id" fetch "https://..."

Automate

Run things on a timer. React when data changes.

every day at 9am digest: save ... every 5 minutes check: run fn/x when orders changes call "https://..." schedules webhooks

Control Access

Control who can read and write your data.

share myapp with @collaborator-key unshare myapp from @collaborator-key

Modes: protected (reads public), shared (reads locked), writeonly (writes public)

Count Things

Atomic add/subtract on any field.

add 1 to stats @views count add 10 to stats @views count subtract 2 from stats @views count

Email

Send emails, manage custom domains.

send email to "[email protected]" subject "Hi" body "..." mail claim "username" mail domains mail domain add "yourdomain.com"

Calendar

Events, RSVP, ICS feeds.

event cal @standup title "Daily" start "..." events cal rsvp cal @standup accepted calendar feed "my-calendar"

Key Management

Rotate keys, manage sub-keys.

key rotate key migrate "old-key" "new-key" subkey read "sub-key-id" subkey update "id" name: "new"

Files

Initiate uploads, list and delete files.

upload "media/photos" "sunset.jpg" blob list "media/photos" blob delete "media/photos" @"sunset.jpg"

Admin

Platform stats, user management, io grants.

admin stats admin users admin logins admin grant "KEY" 1000

Notifications

Send push notifications to browsers.

push "Title" to ("KEY") body "text" push subscribe "route/*" push subscriptions push unsubscribe "id"

Values & Tokens

Create your own currency. Mint, transfer, burn.

value create "gold" name "Gold" symbol "GLD" value mint "gold" 1000 value transfer "gold" 100 to "KEY" value burn "gold" 50 value balance "gold" value supply "gold" value currencies value holders "gold" value ledger "gold"

Validation

Define patterns, validate your data.

define "todo" version "1" fields title: "string" patterns validate "todo" title: "Buy milk"

Smart Fields

Search across all your data by date, location, identifier, amount.

primitives dates "$gte": "2026-01-01" ... identifiers "$has": type: "email"

Chain & Combine

Save many at once. Chain transforms.

bulk (save a @x v: 1, save b @y v: 2) find users then pick name then first 10

SQL

Full SQL when you need it.

sql "SELECT * FROM users WHERE age > 25" sql "INSERT INTO users (name) VALUES ('Alice')"

Account

Check who you are, your balance, your routes.

whoami balance transfer 100 to "KEY" routes context

Export & History

Stream data out. Browse entity versions.

export users export orders latest ledger list users

Security

2FA, multi-sig, GDPR compliance.

totp setup totp verify "123456" multisig propose "route" action: "grant" gdpr erase "route" gdpr redact "route" @key fields: "email"

Domains & Identity

Custom domains, handles, purchasing.

domain add "myapp.com" handle set "myname" resolve "myname" purchase "route" @item

Trust Chain

Verify data integrity. SHA-256 hash chain.

chain status chain verify

Network

Node info, sync status, documentation.

nodes node status sync status docs docs "topic"

Pipes

Chain commands together. Results flow from one step to the next. (The pipe | also works as an alternative.)

findget results

→

then pick nameextract field

→

then sort namereorder

→

then first 5take first

-- Get the 3 most recent error messages
find logs where level = "error" sort created_at desc then pick msg then first 3

-- Count users
count users

-- Get the first order
find orders sort created_at then first

-- Extract names and emails
users where status = "active" then pick name, email

Data transforms: pick, first, last, flat, sort (add asc or desc, default asc), skip, limit, count. (pluck and .field dot syntax still work for backwards compatibility.)

Pipe Actions

Find something, then do something with it:

-- Email the latest customer
find customers sort created_at desc then first then send email to email subject "Welcome!" body "Thanks for joining."

-- Back up the first 10 users
find users then first 10 then save backups/users @snapshot

-- Delete old logs
find old/logs then remove

Action transforms: save route, send email to field, remove. The send email reads the recipient from a field in the piped data. If no body is given, the data itself becomes the email body.

Build a Complete App in 10 Lines

-- 1. Deploy a news fetcher function
save fn/news @main "import urllib.request, json
def main(ocean, request):
  data = json.loads(urllib.request.urlopen('https://api.example.com/news').read())
  for a in data[:10]:
    ocean.write('news/articles', a['id'], a)
  return {'fetched': len(data[:10])}"
save fn/news @_config execute: "public"

-- 2. Run it every hour
every hour news-fetch: run fn/news

-- 3. Get notified when articles arrive
when news/articles changes call "https://hooks.slack.com/..."

-- 4. Publish a dashboard
save _site @news "<html><body><h1>News</h1></body></html>"

-- 5. Lock it down
share fn/news with @your-key mode protected
share news with @your-key mode protected

The Old Way vs. Drops

Task	Drops	The Old Way
Store data	save users @alice name: "Alice"	Multiple tools, custom code, database setup
Find things	find users where role = "admin"	Multiple tools, custom code, database setup
Deploy function	save fn/x @main "def main(o,r): ..."	Complex deployment pipeline
Schedule	every day at 9am daily: run fn/x	Multiple services wired together
Send email	send email to "[email protected]" subject "Hi" body "Hey"	Email service with complex setup
Push notification	push "New!" to ("key") body "Update"	Pages of security configuration
Full-text search	search docs "auth flow"	Separate search service
Access control	share route with @collaborator	Layers of complexity
React to changes	when orders changes call "url"	Layers of complexity
Aggregate	find orders group category sum amount	Multiple tools, custom code, database setup

Syntax Reference

Naming Things

route                     -- letters, digits, _ / -
route @key               -- entity key
route @"[email protected]" -- quoted key for special chars

Adding Information

name: "Alice" age: 30 active: true          -- key:value pairs (preferred)
{ name: "Alice", age: 30, active: true }   -- {} also works
("a", "b", "c")                            -- arrays (parens preferred)
["a", "b", "c"]                            -- [] also works
"string"  42  3.14  true  false  null    -- literals

Finding Things

name = "Alice"      age > 10      tags contains "urgent"
status in ("active", "pending")    email exists
age > 10 and role = "admin"

Field names are bare (no dot prefix needed). Single = for equality.

Shortcuts

users @alice                       -- get (route + @key)
users @alice name: "Alice"        -- save (route + @key + payload)
users where age > 10             -- find (route + where)
users                              -- list (bare route)

Summarizing

find orders group category sum amount count
find users group role count
find users then pick name, email
count users

Scheduling

every day at 9am digest: save reports @daily type: "digest"
every hour check: run fn/health
every monday at 8am weekly: send email to "[email protected]" subject "Week start"
every 5 minutes tick: add 1 to stats @ticks count
every midnight cleanup: remove temp @old
-- Options: minute, hour, day, weekday, midnight, noon, monday-sunday, N minutes, N hours, at 9am, at 5:30pm

Multiple Steps

-- Newlines or semicolons separate commands
save a @x v: 1
a @x

-- Same thing on one line
save a @x v: 1; a @x

-- Comments start with --

Works with AI Agents

Works with AI tools. Connect Claude, Cursor, or any AI assistant — it can build complete apps using Drops:

{
  "mcpServers": {
    "ocean": {
      "url": "https://api.infiniteocean.io/mcp/default",
      "headers": { "X-Ocean-Key": "YOUR_KEY" }
    }
  }
}

The agent gets a single exec tool and can write multi-line Drops commands to save data, deploy functions, set up schedules, send emails, and control access — all in one tool call. It also gets a built-in docs command to read platform documentation.