CML Lab Lifecycle Management

MCP Server

• Command: cml-mcp (pip-installed, stdio transport)
• Requires: CML_URL, CML_USERNAME, CML_PASSWORD environment variables
• Optional: CML_VERIFY_SSL (default: true)

Available Tools

Lab CRUD Operations

| Tool | Parameters | What It Does | |------|-----------|-------------| | create_lab | title | Create a new empty lab with a title | | get_lab | lab_id or lab_title | Get lab details (state, node count, link count) | | get_labs | none | List all labs on the CML server | | start_lab | lab_id or lab_title | Start all nodes in a lab (boot the topology) | | stop_lab | lab_id or lab_title | Stop all nodes in a lab (graceful shutdown) | | wipe_lab | lab_id or lab_title | Wipe lab node configurations (reset to factory) | | delete_lab | lab_id or lab_title | Delete a lab permanently |

Lab Import/Export

| Tool | Parameters | What It Does | |------|-----------|-------------| | import_lab | topology (YAML string) | Import a lab from CML topology YAML | | export_lab | lab_id or lab_title | Export a lab as CML topology YAML | | clone_lab | lab_id or lab_title | Clone an existing lab (deep copy) | | download_lab_configs | lab_id or lab_title | Download all node configs from a running lab |

Workflow: Create a Lab from Scratch (Slack)

When a user says something like "build me a 3-router OSPF lab" in Slack:

1. Create the lab: Use create_lab with a descriptive title
2. Add nodes: (See cml-topology-builder skill) Add routers, switches, servers
3. Wire the topology: (See cml-topology-builder skill) Create interfaces and links
4. Apply configs: (See cml-node-operations skill) Set startup configs
5. Start the lab: Use start_lab to boot everything
6. Verify convergence: (See cml-node-operations skill) Execute show commands to verify routing
7. Report back: Tell the user the lab is ready with node names and management IPs

Example Slack Conversations

"Build me a 3-router OSPF lab" → create_lab "OSPF Lab - 3 Routers" → create_node x3 (IOSv: R1, R2, R3) → create_interface + create_link (R1-R2, R2-R3, R1-R3 full mesh) → set_node_config for each (OSPF area 0, loopbacks, point-to-point links) → start_lab → execute_command: "show ip ospf neighbor" on each router → Report: "Lab ready! R1, R2, R3 all have FULL OSPF adjacencies."

"Add a 4th router to the OSPF lab" → stop_lab (topology changes require stopped state) → create_node R4 (IOSv) → create_interface + create_link (R4 to R2 and R3) → set_node_config for R4 (OSPF area 0) → Update R2 and R3 configs if needed → start_lab → Verify all 4 routers have adjacencies

"Create a lab from this YAML topology" → import_lab with the provided YAML → start_lab → Report node count, link count, state

"Clone the OSPF lab and add BGP peering" → clone_lab "OSPF Lab - 3 Routers" → "OSPF+BGP Lab" → Modify configs to add BGP router stanzas → start_lab → Verify both OSPF and BGP adjacencies

Workflow: Lab Cleanup

When a user is done with a lab:

1. Stop the lab: stop_lab — graceful shutdown of all nodes
2. Optional export: Ask if they want to save the topology (export_lab)
3. Wipe configs: wipe_lab to reset node configs (if reusing later)
4. Delete the lab: delete_lab to free CML resources

Workflow: Clone and Modify

When a user wants to experiment with a variant of an existing lab:

1. Clone the lab: clone_lab creates a deep copy
2. Modify the clone: Add/remove nodes, change links, update configs
3. Start the clone: start_lab on the new copy
4. Keep the original untouched for comparison

Workflow: Export for Sharing

When a user wants to share a lab topology:

1. Export: export_lab returns CML YAML topology
2. Optionally commit to GitHub: Use github-ops skill to commit the YAML to a repo
3. Share via Slack: Post the YAML or share the GitHub link

Lab States

Labs go through these states:

| State | Meaning | |-------|---------| | DEFINED_ON_CORE | Lab exists but no nodes are running | | STARTED | All (or some) nodes are booting/running | | STOPPED | All nodes have been shut down | | BOOTED | All nodes have finished booting |

Common Lab Patterns

CCIE Routing Lab

"Create a CCIE routing lab with:
- 4 x IOSv routers (R1-R4)
- 2 x IOSv L2 switches (SW1-SW2)
- Full mesh between R1-R4
- SW1 connecting R1 and R2, SW2 connecting R3 and R4"

Data Center Lab

"Build a DC lab:
- 2 x NX-OS spines
- 4 x NX-OS leaves
- 1 x external router (IOSv)
- Spine-leaf full mesh, external router to both spines"

Service Provider Lab

"Create an SP lab:
- 4 x IOS-XR routers (PE1, PE2, P1, P2)
- 2 x IOSv CE routers
- PE-P full mesh, CE1 to PE1, CE2 to PE2"

Important Rules

• Always check lab state before operations — don't try to start an already-running lab
• Use get_labs first to see what already exists — avoid duplicate lab names
• Export before delete — ask the user if they want to save the topology before destroying it
• Lab titles can be used instead of lab IDs — the CML MCP resolves titles to IDs
• Wipe requires stop — a lab must be stopped before configs can be wiped
• Record in GAIT — log lab creation/deletion for audit trail

Environment Variables

• CML_URL — CML server URL (e.g., https://cml.example.com)
• CML_USERNAME — CML username
• CML_PASSWORD — CML password
• CML_VERIFY_SSL — Verify SSL certificate (true/false)