Scripted automation (no LLM)

Not everything needs an agent. The simplest use of the SDK is as a typed API client: resolve a connection, execute a tool, use the output.

from engini import Engini

client = Engini()  # reads ENGINI_API_KEY
conn_id = next(c.connection_id for c in client.connections.list(application="salesforce"))

result = client.tools.execute(
    "salesforce_getrecords", {"sobject": "Account"}, connection_id=conn_id
)
accounts = result.output or []

print(f"{len(accounts)} accounts (history {result.history_id}):")
for account in accounts[:10]:
    print(f"  - {account.get('Name')}  [{account.get('Id')}]")

That’s a complete export script. Wrap it in your scheduler of choice, add typed error handling for retries on EnginiRateLimitError, and you have a production sync. The same thing from the CLI:

engini tools call salesforce_getrecords --args '{"sobject":"Account"}' --connection 3 \
  | jq '.output | length'

Source: python/examples/salesforce_export.py (uv run salesforce_export.py).

⌘I

Start here

REST API

CLI

SDK