Integration administrator or developer
Connect a provider only after proving its trust path
A connector record says how one workspace reaches another system. Production support requires more than a successful setup screen: scopes, refresh behavior, action and trigger proof, permission-denied behavior, rate-limit handling, and evidence must all be understood.
Before you start
- Provider sandbox or test account.
- Approved OAuth client or least-privilege credential.
- Expected actions, events, and negative cases documented.
#Start in the marketplace
Choose a native connector when its supported operations match the business need. Use a typed custom REST operation when no native action exists. Read readiness labels and required provider IDs before creating the connector.
#Configure and test least privilege
- Select the intended workspace and shared provider account.
- Enter encrypted credentials or complete OAuth.
- Grant only scopes required by the selected operations.
- Run health and a harmless sandbox action.
- Verify an inbound event separately, including signature and timestamp.
- Test expired credentials, denied permission, and provider throttling.
#Keep claims tied to evidence
Use the certification page to record provider setup, scope review, action samples, trigger samples, token refresh, denied permission, rate-limit behavior, and sanitized screenshots or logs. Mark “Sandbox verified” only when the checklist is complete.
#Common mistakes
- Using a personal provider account for shared work.
- Granting broad scopes for convenience.
- Calling setup completion production certification.
#Verify the result
- Credentials are encrypted and least-privilege.
- Action and inbound event paths were tested separately.
- Refresh and negative cases are recorded.
- The readiness label matches evidence.