Summary
Deploy, then debug the live pod, in the same conversation.
The debugging half is what separates this from a deploy button. An agent that just shipped a broken build can open a shell in the pod, read the file, edit it in place and restart — instead of a full redeploy cycle between every guess.
What it is
A hosted server at https://mcp.nexlayer.ai/api/mcp that hands a coding agent a production Kubernetes platform: deploy from one nexlayer.yaml, then read logs, open a shell in a pod, edit a file in place, or scale it.
What you get
- Deployment: nexlayer_validate_yaml before it, nexlayer_deploy, then nexlayer_check_deployment_status, nexlayer_get_deployment_logs and nexlayer_get_deployment_events when something fails
- Live debugging: nexlayer_debug_shell_open and send, nexlayer_debug_file_list, copy in and out, and nexlayer_debug_file_edit for an in-place change with no redeploy
- Pods and network: describe, restart, scale, plus nexlayer_debug_proxy_http, nexlayer_debug_namespace_dns and nexlayer_debug_db_query against a database pod
- Custom domains: nexlayer_get_domain_setup_guide, nexlayer_add_domain_to_profile, nexlayer_check_domain_configuration and nameserver verification
- Account: API keys created, listed and revoked, plus nexlayer_check_credits, purchase, plan upgrade and coupons
- Prebuilt agent templates deployed in one call — chatbot, code copilot, code reviewer, content guard, multi-agent and translation hub
Requirements
A Nexlayer account and one YAML file describing the deployment. Connect with npx @nexlayer/mcp-install, or add the endpoint to your editor's MCP config by hand. Billing runs on credits, and the agent can read the balance itself.
Setup effort
One command — npx @nexlayer/mcp-install