Triggering a deploy is easy; the useful part is that when it fails the assistant can pull the full log and tell you what broke without you opening the dashboard. Preview mode on create_deployment is worth knowing about — it lets you see what a commit range would do before queueing it for real.
A DeployHQ server covering projects, servers and deployments, plus the account-level pieces deployments depend on: SSH keys, global environment variables and config file templates.
- list_projects and get_project — projects with their repository information and deployment status
- list_servers — the servers configured for a project
- list_deployments and get_deployment — deployments with pagination and per-server filtering, and one deployment in detail
- get_deployment_log — the complete deployment log as text, which is where a failure explains itself
- create_deployment — deploy a commit range to a server or server group, with options for branch, queue or preview mode, config file copying, build commands and the build cache
- list_ssh_keys and create_ssh_key — account SSH keys, listed with fingerprints and never returning private keys
- list_global_environment_variables, create_global_environment_variable, update_global_environment_variable, delete_global_environment_variable — account-wide variables, with locking and build-pipeline availability
- list_global_config_files, get_global_config_file, create_global_config_file, update_global_config_file, delete_global_config_file — config file templates shared across projects
A DeployHQ account. The published server definition takes DEPLOYHQ_EMAIL, DEPLOYHQ_API_KEY and DEPLOYHQ_ACCOUNT — the account name from your DeployHQ URL — with LOG_LEVEL optional. Installed from npm as deployhq-mcp-server and run over stdio with npx, so there is nothing to install first. SSE and HTTP transports are available for hosting it.
One command plus a key — npx -y deployhq-mcp-server, then supply credentials