Verify With Canaries Before Cutover

verify-with-canaries-before-cutover

Summary

Automation is risky when production cutover is the first real proof. Use a canary path, rollback plan, and human approval for operations with remote access, credentials, production data, or long-lived service impact.

Problem

Automation is risky when the first real test is the production cutover. Agents can confidently change a system while missing reachability, rollback, credential, or compatibility problems that only appear after deployment.

Preferred Move

For risky operations, create a canary path that proves the important property before production cutover. Validate the canary, preserve a rollback path, and require human approval when the blast radius or irreversibility warrants it.

Tradeoff

Canaries add overhead. Skip them for low-risk local changes with cheap rollback, but prefer them when the operation touches remote access, credentials, devices, production data, or long-lived services.

Agent Instruction

Before cutting over a risky automation, identify the canary signal and rollback path. Do not treat
the cutover as complete until the canary evidence is reported.

Examples

Bad: the agent replaces the only working access path in one step.

Uploaded the new remote-access binary and restarted the service.

Good: the agent tests a separate identity before production cutover.

Deployed the new binary to the canary identity, verified SSH reachability, then cut over the
production identity after the check passed.