ServiceNow Change Workflow

Golden Rule

NEVER execute a network change without an approved Change Request. The only exception is an Emergency change, which still requires a CR and immediate human notification.

How to Call the Tools

The ServiceNow MCP server provides change management and incident tools. Call them via mcp-call:

python3 $MCP_CALL "python3 -u $SERVICENOW_MCP_SCRIPT" TOOL_NAME '{"param":"value"}'

ServiceNow Change Request States

New -> Assess -> Authorize -> Scheduled -> Implement -> Review -> Closed -> Canceled

• New: CR created, not yet submitted

• Assess: Submitted for review / risk assessment

• Authorize: Awaiting approval

• Scheduled: Approved and scheduled for implementation window

• Implement: Approved and ready for execution (this is the gate)

• Review: Post-implementation review

• Closed: Change completed and verified

Change Types

Type Approval Use When

Normal CAB / manager approval required Interface config, routing changes, ACL updates, any production change

Standard Pre-approved template Password rotation, NTP update, banner change

Emergency Expedited (post-facto approval OK) Active outage, security incident, critical vulnerability

Full Change Lifecycle

Phase 0: Pre-Change Incident Check

Before creating a CR, verify no open P1/P2 incidents exist on affected CIs. Executing changes during active incidents violates ITIL best practice and risks compounding the outage.

0A: Check for Open Critical Incidents

python3 $MCP_CALL "python3 -u $SERVICENOW_MCP_SCRIPT" list_incidents '{"limit":20,"state":"1","query":"priority=1^ORpriority=2"}'

State 1 = New/Open. Review results for any incidents affecting the target device or service.

Decision gate:

• Open P1/P2 on affected CI -> STOP. Do not proceed. Notify the human.

• Open P1/P2 on unrelated CI -> Proceed with caution, note in CR description.

• No open P1/P2 -> Proceed.

0B: Check for Active Change Freezes

python3 $MCP_CALL "python3 -u $SERVICENOW_MCP_SCRIPT" list_change_requests '{"limit":10,"state":"implement","type":"normal"}'

Review active in-progress changes for conflicts. Two changes on the same device at the same time is a collision.

Phase 1: Create Change Request

1A: Create the CR

python3 $MCP_CALL "python3 -u $SERVICENOW_MCP_SCRIPT" create_change_request '{"short_description":"Add Loopback99 to R1 for OSPF RID migration","description":"Add Loopback99 (99.99.99.99/32) to R1 as new OSPF router-id. Pre-change baseline captured. Rollback plan: no interface Loopback99. Affected CI: R1 (devnetsandboxiosxec8k.cisco.com). No open P1/P2 incidents on this CI.","type":"normal","category":"Network","risk":"moderate","impact":"3","assignment_group":"Network Engineering"}'

CR description must include:

• What is being changed and why

• Affected CI (device hostname, management IP)

• Pre-change incident status (from Phase 0)

• Rollback plan

• Expected impact and duration

• Verification criteria

1B: Add Change Tasks

Break the change into discrete tasks for tracking:

python3 $MCP_CALL "python3 -u $SERVICENOW_MCP_SCRIPT" add_change_task '{"change_id":"CHG0000123","short_description":"Capture pre-change baseline on R1","description":"Save running config, OSPF neighbors, routing table, interface states, and connectivity baselines."}'

python3 $MCP_CALL "python3 -u $SERVICENOW_MCP_SCRIPT" add_change_task '{"change_id":"CHG0000123","short_description":"Apply Loopback99 configuration on R1","description":"Apply: interface Loopback99, ip address 99.99.99.99 255.255.255.255, description OSPF-RID-Migration, no shutdown"}'

python3 $MCP_CALL "python3 -u $SERVICENOW_MCP_SCRIPT" add_change_task '{"change_id":"CHG0000123","short_description":"Post-change verification on R1","description":"Verify Loopback99 up/up, OSPF neighbors stable, routing table +1 connected route, connectivity 100%."}'

1C: Record CR Creation in GAIT

python3 $MCP_CALL "python3 -u $GAIT_MCP_SCRIPT" gait_record_turn '{"user_text":"Create change request for Loopback99 addition on R1","assistant_text":"Created CR CHG0000123: Add Loopback99 to R1 for OSPF RID migration. Type: normal, Risk: moderate, Impact: 3. 3 change tasks added. Pre-change incident check: CLEAR (no P1/P2 on R1)."}'

Phase 2: Approval Gate

2A: Submit for Approval

python3 $MCP_CALL "python3 -u $SERVICENOW_MCP_SCRIPT" submit_change_for_approval '{"change_id":"CHG0000123","approval_comments":"Pre-change checks complete. No open P1/P2 incidents. Rollback plan documented. Ready for CAB review."}'

2B: Wait for Approval

Poll the CR status until it reaches Implement state:

python3 $MCP_CALL "python3 -u $SERVICENOW_MCP_SCRIPT" get_change_request_details '{"change_id":"CHG0000123"}'

Decision gate:

• State = implement -> Proceed to Phase 3

• State = authorize -> Still awaiting approval. Wait.

• State = canceled -> CR was rejected. STOP. Notify the human with rejection reason.

CRITICAL: Do NOT proceed to Phase 3 unless the CR state is implement (approved).

2C: Approve Change (when authorized to do so)

If the human operator has approval authority:

python3 $MCP_CALL "python3 -u $SERVICENOW_MCP_SCRIPT" approve_change '{"change_id":"CHG0000123","approval_comments":"Approved. Rollback plan verified. Maintenance window confirmed."}'

Phase 3: Execution

This phase uses the pyats-config-mgmt skill for the actual device work. The ServiceNow CR gates entry to this phase.

3A: Update CR to Implementation

python3 $MCP_CALL "python3 -u $SERVICENOW_MCP_SCRIPT" update_change_request '{"change_id":"CHG0000123","work_notes":"Beginning implementation. Capturing pre-change baseline."}'

3B: Pre-Change Baseline (via pyats-config-mgmt)

Follow the pyats-config-mgmt skill Phase 1 procedure:

PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "python3 -u $PYATS_MCP_SCRIPT" pyats_show_running_config '{"device_name":"R1"}'

PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "python3 -u $PYATS_MCP_SCRIPT" pyats_run_show_command '{"device_name":"R1","command":"show ip ospf neighbor"}'

PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "python3 -u $PYATS_MCP_SCRIPT" pyats_run_show_command '{"device_name":"R1","command":"show ip route"}'

PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "python3 -u $PYATS_MCP_SCRIPT" pyats_ping_from_network_device '{"device_name":"R1","command":"ping 8.8.8.8 repeat 10"}'

Update the CR with baseline status:

python3 $MCP_CALL "python3 -u $SERVICENOW_MCP_SCRIPT" update_change_request '{"change_id":"CHG0000123","work_notes":"Pre-change baseline captured: Running config saved, 2 OSPF neighbors FULL, 47 routes, 100% connectivity to 8.8.8.8 (23ms RTT)."}'

3C: Apply Configuration (via pyats-config-mgmt)

PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "python3 -u $PYATS_MCP_SCRIPT" pyats_configure_device '{"device_name":"R1","config_commands":["interface Loopback99","ip address 99.99.99.99 255.255.255.255","description OSPF-RID-Migration","no shutdown"]}'

Update the CR:

python3 $MCP_CALL "python3 -u $SERVICENOW_MCP_SCRIPT" update_change_request '{"change_id":"CHG0000123","work_notes":"Configuration applied to R1. Proceeding to post-change verification."}'

Phase 4: Post-Change Verification

4A: Verify the Change (via pyats-config-mgmt)

Follow the pyats-config-mgmt skill Phase 4 procedure:

PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "python3 -u $PYATS_MCP_SCRIPT" pyats_show_running_config '{"device_name":"R1"}'

PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "python3 -u $PYATS_MCP_SCRIPT" pyats_run_show_command '{"device_name":"R1","command":"show ip ospf neighbor"}'

PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "python3 -u $PYATS_MCP_SCRIPT" pyats_run_show_command '{"device_name":"R1","command":"show ip route"}'

PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "python3 -u $PYATS_MCP_SCRIPT" pyats_ping_from_network_device '{"device_name":"R1","command":"ping 8.8.8.8 repeat 10"}'

PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "python3 -u $PYATS_MCP_SCRIPT" pyats_show_logging '{"device_name":"R1"}'

4B: Verification Decision

Compare pre-change vs post-change state:

Check Pre-Change Post-Change Expected Delta

OSPF neighbors 2 FULL 2 FULL No change

Route count 47 48 +1 connected (99.99.99.99/32)

Connectivity 100% 100% No change

New errors in log

%LINEPROTO-5-UPDOWN Lo99 up Expected

Decision gate:

• All checks PASS -> Proceed to Phase 5 (Close)

• Any check FAILS -> Initiate Rollback (Phase 4C), do NOT close the CR

4C: Rollback (If Verification Fails)

If post-change verification fails, roll back immediately:

PYATS_TESTBED_PATH=$PYATS_TESTBED_PATH python3 $MCP_CALL "python3 -u $PYATS_MCP_SCRIPT" pyats_configure_device '{"device_name":"R1","config_commands":["no interface Loopback99"]}'

Update the CR with rollback details:

python3 $MCP_CALL "python3 -u $SERVICENOW_MCP_SCRIPT" update_change_request '{"change_id":"CHG0000123","work_notes":"ROLLBACK EXECUTED: Post-change verification failed. [Describe what failed]. Configuration rolled back. Device restored to baseline state. Verification of rollback: [results]."}'

Create an incident for the failed change:

python3 $MCP_CALL "python3 -u $SERVICENOW_MCP_SCRIPT" create_incident '{"short_description":"Failed change CHG0000123 on R1 - rollback executed","description":"Change CHG0000123 (Add Loopback99 to R1) failed post-change verification. Rollback executed successfully. Device restored to pre-change state. Root cause investigation required.","urgency":"2","impact":"2","category":"Network"}'

CRITICAL: If rollback fails, this becomes a P1 incident. Notify the human IMMEDIATELY.

Phase 5: Close the Change Request

5A: Update CR with Final Status

python3 $MCP_CALL "python3 -u $SERVICENOW_MCP_SCRIPT" update_change_request '{"change_id":"CHG0000123","work_notes":"Post-change verification PASSED. OSPF neighbors stable (2 FULL), route table +1 connected (99.99.99.99/32), connectivity 100%, no unexpected log entries. Change successful.","state":"review"}'

5B: Close the CR

python3 $MCP_CALL "python3 -u $SERVICENOW_MCP_SCRIPT" update_change_request '{"change_id":"CHG0000123","state":"closed","work_notes":"Change completed successfully. All verification checks passed. No rollback required."}'

5C: Record Full Session in GAIT

python3 $MCP_CALL "python3 -u $GAIT_MCP_SCRIPT" gait_record_turn '{"user_text":"Execute change CHG0000123: Add Loopback99 to R1","assistant_text":"Change CHG0000123 completed successfully.\n\nTimeline:\n- Pre-check: No open P1/P2 incidents on R1\n- CR created and approved\n- Baseline: 2 OSPF neighbors FULL, 47 routes, 100% connectivity\n- Config applied: interface Loopback99, 99.99.99.99/32\n- Verification: PASSED (2 OSPF FULL, 48 routes, 100% connectivity)\n- CR closed: successful\n\nArtifacts: Pre/post running config diff, routing table diff, log entries."}'

5D: Display GAIT Audit Trail

python3 $MCP_CALL "python3 -u $GAIT_MCP_SCRIPT" gait_show '{"commit":"HEAD"}'

Emergency Change Process

Emergency changes bypass the normal approval gate but still require a CR and immediate human notification.

When to Use Emergency Changes

• Active outage requiring immediate remediation

• Security incident requiring immediate mitigation (e.g., ACL to block attack)

• Critical vulnerability with active exploitation

Emergency Workflow

E1: Create Emergency CR

python3 $MCP_CALL "python3 -u $SERVICENOW_MCP_SCRIPT" create_change_request '{"short_description":"EMERGENCY: Block attack source 203.0.113.99 on R1","description":"EMERGENCY CHANGE: Active DDoS from 203.0.113.99 detected. Applying inbound ACL on GigabitEthernet1 to block source. Post-facto approval required.","type":"emergency","category":"Network","risk":"high","impact":"1","assignment_group":"Network Engineering"}'

E2: Notify Human Immediately

CRITICAL: For emergency changes, you MUST notify the human operator before executing. State:

• The emergency condition

• The proposed remediation

• The CR number

• Request verbal/written approval to proceed

E3: Execute Upon Human Approval

Follow Phase 3 (Execution) with accelerated timelines. Baseline capture is still mandatory.

E4: Post-Facto Approval

After execution, submit the emergency CR for post-facto approval:

python3 $MCP_CALL "python3 -u $SERVICENOW_MCP_SCRIPT" submit_change_for_approval '{"change_id":"CHG0000456","approval_comments":"EMERGENCY CHANGE - Post-facto approval requested. Change executed during active incident. Full audit trail in GAIT."}'

Integration with Other Skills

Skill Integration Point

pyats-config-mgmt Phase 3 (Execution) and Phase 4 (Verification) - actual device configuration

pyats-health-check Pre-change device health validation before entering Phase 3

netbox-reconcile Post-change: verify NetBox source of truth reflects the change (or ticket the delta)

GAIT Every phase records to the audit trail - CR creation, approval, execution, verification, closure

markmap-viz Generate change summary mind map for complex multi-device changes

Change Report Format

After every change, produce this summary:

Change Report - CHG0000123 Date: YYYY-MM-DD HH:MM UTC Device: R1 (devnetsandboxiosxec8k.cisco.com) Type: Normal | Risk: Moderate | Impact: 3

Pre-Check: No open P1/P2 incidents on R1 Approval: Approved by [approver] at HH:MM UTC

Pre-Change State:

• OSPF neighbors: 2 (FULL)
• Routes: 47
• Connectivity: 100% to 8.8.8.8 (23ms)

Config Applied: interface Loopback99 ip address 99.99.99.99 255.255.255.255 description OSPF-RID-Migration no shutdown

Post-Change State:

• OSPF neighbors: 2 (FULL) - no change
• Routes: 48 (+1 connected 99.99.99.99/32)
• Connectivity: 100% to 8.8.8.8 (23ms) - no change
• Log: %LINEPROTO-5-UPDOWN Lo99 up/up (expected)

Verification: PASSED Rollback Required: No CR Status: Closed (successful) GAIT Commit: [commit hash]