What rollout checks should we complete before turning on new integrations or skills?

Checklist before enabling integrations, MCP connectors, or new skills—including Endpoint AI review for connectors and package skills when your organization uses it.

Use this checklist before you flip on a new integration, MCP connector, or skill for all users—especially when Endpoint AI governs review and endpoint rollout.

Identity and access

SSO / login paths tested for the pilot group you will expand.
Roles in Harriet match who may configure versus consume integrations and skills.
Review queue: confirm who may submit, approve, or request changes for connectors and package skills (submitter and reviewer should not be the same person unless policy allows).

Data and compliance

Data residency and DPA coverage confirmed for the vendor you connect.
Least privilege OAuth scopes or API keys—remove unused access.

Operational readiness

Runbook for credential rotation and who owns it after go-live.
Monitoring: who watches for failed syncs, rate limits, or auth expiry.
Endpoint AI: after Sync tools on a new MCP connector, verify tool permissions and sandbox behavior match your risk appetite.

User communication

Short employee-facing note: what changed, what they can ask, and what to do if something fails (open a ticket).

Example

Before enabling expenses, finance signs off on mapping, IT verifies sandbox claims, HR trains champions, then you enable the skill for one country before global.

Guardrails

Do not enable write skills in production until read paths are stable.
Keep a pilot cohort long enough to see at least one full pay cycle or equivalent business rhythm if payroll or HRIS is involved.
Package skills submitted via Skilify are not auto-published—plan reviewer capacity before inviting broad submission.

Related: How to create an MCP connector (Company settings), MCP from GitHub: sandboxed connectors vs skill packages, and Using Skilify to build skills locally, submit for review, and provision them.