The Problem
AI coding assistants such as Claude, Cursor, TRAE, Coze, and SOLO are powerful, but they all share a fundamental limitation: they run inside sandboxed cloud environments with no direct SSH access to your servers.
Potencialize seu agente
Instalação do publicador. Não é um endpoint do PlugMCP.
Host a copy with our keys and logs.
Ainda não há uma versão empacotada no registro. Siga o README do repositório.
URL remota ou um comando na sua máquina? Como as duas instalações diferem
Rodar no PlugMCP
A new server you create. It is not this listing's official endpoint. Use the publisher install when you want their URL and their credentials.
AbrirDo publicador
AI coding assistants such as Claude, Cursor, TRAE, Coze, and SOLO are powerful, but they all share a fundamental limitation: they run inside sandboxed cloud environments with no direct SSH access to your servers.
The AI-native deployment gateway — bridge sandboxed AI IDEs to your infrastructure via MCP protocol
This means your AI assistant cannot:
The situation is even harder in enterprise environments where servers sit behind bastion hosts or jump servers that cloud-based IDEs cannot reach at all. Manual deployment remains tedious, error-prone, and wastes developer time.
DeployPilot bridges this gap by running on your server and exposing a standard MCP (Model Context Protocol) interface that any AI IDE can call — securely, remotely, and autonomously.
DeployPilot is a self-hosted AI IDE deployment gateway that uses the MCP protocol to let sandboxed AI assistants manage your server infrastructure. It installs in one line and gives your AI IDE full control over deployment, DNS, SSL, monitoring, and more.
One command to install. One prompt to deploy.
docker compose up -d
Then tell your AI:
"Deploy my project, set up DNS, and provision SSL."
DeployPilot handles the rest.
# Clone the repository
git clone https://github.com/Yogdunana/deploypilot.git
cd deploypilot
# Set required environment variables
export JWT_SECRET=$(openssl rand -base64 24)
export REDIS_PASSWORD=$(openssl rand -base64 24)
# Start the service
docker compose up -d
Open http://localhost:8080 and register an admin account.
curl -fsSL https://raw.githubusercontent.com/Yogdunana/deploypilot/main/scripts/install.sh | bash
The install script automatically downloads the latest binary, generates admin credentials, and sets up systemd services.
git clone https://github.com/Yogdunana/deploypilot.git
cd deploypilot && make build-all
See DEPLOY.md for the full deployment guide, including PostgreSQL setup, reverse proxy configuration, and production hardening.
DeployPilot uses a three-layer architecture designed for both AI agents and human operators:
flowchart LR
subgraph S1["AI IDE Sandbox"]
A[Claude / Cursor / TRAE / Coze / SOLO]
end
subgraph S2["DeployPilot Gateway"]
B["MCP Server - 91+ tools"]
C["REST API - JWT + RBAC"]
D[WebSocket / SSE]
E[Deploy Engine]
F[Provider Plugins]
end
subgraph S3["Infrastructure"]
G["1Panel / BT Panel"]
H["Docker / Kubernetes"]
I["Cloudflare / Aliyun DNS"]
J["GitHub Actions / Gitea"]
end
A --> B
A --> C
C --> E
B --> E
E --> F
F --> G
F --> H
F --> I
F --> J
The embedded web dashboard (Vue 3 + TypeScript + Tailwind CSS) provides a full management UI accessible at http://localhost:8080.
91+ MCP tools covering the complete deployment lifecycle. AI IDEs connect via stdio transport and can autonomously manage your infrastructure through natural language.
Register and manage multiple servers. Supports SSH-based remote servers, Docker (local and remote), and Kubernetes multi-cluster deployments. Detects server environments and installed panels automatically.
Full container management — create, deploy, start, stop, remove, and monitor containers. Supports automatic port allocation, health checks with configurable retries, backup and rollback, and self-healing with auto-restart and auto-rollback on failure thresholds.
Unified DNS record management across multiple providers:
Request, list, renew, and delete SSL certificates directly from the MCP interface or web dashboard.
Trigger and monitor CI/CD builds without leaving your AI IDE. Supports GitHub Actions and Gitea.
Real-time container and system metrics, configurable alert rules, and automatic container healing when failures are detected.
Sign in with your GitHub or Gitee account — no separate password needed.
Four-tier role-based access control: owner > admin > dev > viewer. Fine-grained permissions on every resource.
Full change tracking with user identity, action type, IP address, and timestamp. Every operation is recorded and auditable.
Web dashboard available in English and Chinese, with a pluggable locale system.
DeployPilot integrates with any IDE that supports the MCP protocol. Configuration details for each IDE are available in docs/ide-skills.md.
All IDEs use the same configuration pattern:
{
"mcpServers": {
"deploypilot": {
"command": "/path/to/mcp-server",
"args": ["--config", "/path/to/config.yaml"]
}
}
}
See docs/ide-skills.md for IDE-specific instructions, skill files, and rules templates.
DeployPilot is configured via config.yaml. All settings can also be overridden with environment variables using the DEPLOYPILOT_ prefix (e.g., DEPLOYPILOT_SERVER_PORT for server.port).
server:
host: "0.0.0.0"
port: 8080
cors_allowed_origins:
- []
database:
type: sqlite # sqlite | postgres
dsn: ./data/deploypilot.db
redis:
addr: localhost:6379
password: ""
db: 0
auth:
jwt_secret: "change-me-in-production" # Required: set a strong random value
token_expire: 24h
deploy:
default_mode: api
build_timeout: 10m
health_check_interval: 30s
health_check_retries: 3
rollback_on_failure: true
log:
level: info # debug | info | warn | error
format: json # json | text
file: ./logs/deploypilot.log
enable_tracing: true
monitor:
enabled: true
metrics_port: 9091
See configs/config.yaml.example for the full example and DEPLOY.md for all environment variables.
Contributions are welcome! Please read CONTRIBUTING.md for development setup, coding conventions, and pull request guidelines.
This project is licensed under the Business Source License 1.1 (BSL 1.1).
> Non-commercial use is permitted — including personal use, internal use within organizations, testing/evaluation, and non-commercial educational use. See LICENSE for full terms. > > On 2029-04-28, this license automatically converts to MIT.
© 2026 Yogdunana
The AI-native deployment gateway — bridge sandboxed AI IDEs to your infrastructure via MCP protocol
Ferramentas
Ainda não há lista de ferramentas. Servidores remotos recebem uma após a primeira verificação de saúde.
Perguntas frequentes
Copie nesta página a configuração stdio para Cursor, Claude Desktop, Claude Code ou VS Code. Ela usa a instalação do publicador, não um proxy do PlugMCP.
Este registro não lista nenhuma variável de ambiente obrigatória.
Deploypilot roda localmente via stdio.
Última verificação: 2026-10-28. The AI-native deployment gateway — bridge sandboxed AI IDEs to your infrastructure via MCP protocol