deployxa_list_projects
Returns all apps in your workspace, branch mappings, active URLs, domains, and last deployment status.
- Inspect active projects & environments
Control deployments directly from Claude, Cursor & AI agents.
The official Deployxa MCP server connects your AI coding assistant to your cloud infrastructure. Inspect builds, stream live logs, evaluate readiness scores, manage secrets, and trigger production rollouts without leaving your editor.
01 / Hosted Endpoint
Production Streamable HTTP gateway.
No local daemons or proxy binaries to install. Connect your editor directly to Deployxa's high-availability hosted MCP endpoint over TLS.
Hosted MCP Gateway
Streamable HTTP · Server-Sent Events · OAuth 2.1 PKCE
Server Endpoint URL
https://mcp.deployxa.com/mcpProtocol Discovery
https://mcp.deployxa.com/.well-known/mcp.jsonSupported AuthenticationOAuth 2.1 PKCE or API Key (
Bearer dxa_...)02 / Quick Setup
Connect in 2 minutes with a copy-paste prompt.
Paste this setup instruction directly into your AI coding assistant. The agent reads its local config, registers the Deployxa MCP endpoint, and verifies access.
One-Prompt Setup
Zero manual JSON editing required.
Modern coding agents can configure their own tool providers. Simply copy the prompt, send it to your assistant, and let it configure the endpoint.
- Works seamlessly in Claude Desktop, Cursor, Windsurf, and Claude Code
- Interactive OAuth browser login or instant API key header
- Immediately unlocks 10+ deployment and diagnostic tools
Agent Instructions
Connect to the Deployxa MCP server at https://mcp.deployxa.com/mcp.
If you support OAuth 2.1, authenticate interactively.
Otherwise, prompt me for a Deployxa API key (starting with dxa_) and configure:
{
"mcpServers": {
"deployxa": {
"url": "https://mcp.deployxa.com/mcp",
"headers": {
"Authorization": "Bearer YOUR_DEPLOYXA_API_KEY"
}
}
}
}
Once connected, call deployxa_list_projects to verify access.03 / Client Guides
Step-by-step configuration for every assistant.
Pick your favorite AI assistant to see the exact configuration snippet and file path.
Config Snippet
{
"mcpServers": {
"deployxa": {
"url": "https://mcp.deployxa.com/mcp"
}
}
}Claude Desktop opens your browser to log in via OAuth 2.1 PKCE on first connection.
04 / Tool Catalog
Everything your agent needs to operate production.
From inspecting project health to rolling back failed releases, Deployxa exposes a rich set of granular, idempotent tools.
deployxa_get_project
Retrieves environment variables, attached Postgres/MySQL databases, build commands, and live metrics.
- Detailed service config & health
deployxa_create_deployment
Deploys a branch, tag, or commit. Automatically triggers container builds, health probes, and edge routing.
- Trigger zero-downtime releases
deployxa_get_deployment_status
Streams status transitions: queued, building, containerizing, running health checks, and live traffic promotion.
- Poll rollout phases in real-time
deployxa_stream_build_logs
Fetches build container stdout/stderr. AI agents analyze missing dependencies, syntax bugs, and lockfile mismatches.
- Diagnose compiler & bundle errors
deployxa_stream_runtime_logs
Streams live container output, uncaught exceptions, 500 status spikes, and database connection timeouts.
- Inspect live application logs
deployxa_rollback_deployment
Re-routes edge traffic to the last known-good immutable container image without recompiling source code.
- Instant one-click rollback
deployxa_get_readiness_score
Calculates project health score, verifies Dockerfile security, detects missing environment secrets, and port conflicts.
- Pre-flight deployment readiness audit
deployxa_set_secret
Adds or updates sensitive API keys, connection strings, and tokens without exposing secrets in raw git commits.
- Manage encrypted environment variables
deployxa_attach_database
Spins up persistent managed database instances and automatically binds connection URLs into runtime environment.
- Provision Postgres, MySQL & Redis
05 / Safety & Governance
Engineered with enterprise safety controls.
AI agents should never have unconstrained power over production. Deployxa includes built-in guardrails and safety checks.
Read-Only Safe Mode
Enforce safe read-only access. When mutation protection is active, agents can inspect logs and readiness scores, but cannot delete projects, modify secrets, or trigger releases.
Scoped Tokens
Issue API tokens restricted to specific workspaces, repositories, or environments. Prevent cross-project leakage and restrict staging keys from accessing production clusters.
Audit Logging
Every tool call, parameter, and execution result initiated by an MCP agent is permanently logged in the Deployxa console with exact timestamps and identity attribution.
06 / Workflow Comparison
Why MCP is the future of developer operations.
Compare the autonomous MCP workflow with manual dashboard navigation and terminal CLI typing.
| Capability | Deployxa MCP Server | Web Dashboard | Terminal CLI |
|---|---|---|---|
| AI Agent Direct Access | Native MCP tool calling | Copy-paste browser UI | Subprocess wrapper only |
| Build Error Auto-Diagnosis | Agent reads logs & fixes code | Manual log inspection | Manual terminal reading |
| One-Click Instant Rollback | Natural language prompt | 3 clicks in UI | dxa rollback command |
| Transport & Protocol | Streamable HTTP / SSE | REST / GraphQL HTTPS | Local CLI binary |
| Safety Guardrails | Safe mode & read-only toggle | Role-based access | User token permissions |
| OAuth 2.1 PKCE Flow | Automatic browser auth | Browser session | dxa login command |
07 / Frequently Asked Questions
Everything you need to know about Deployxa MCP.
What is the Deployxa MCP Server?
The Deployxa Model Context Protocol (MCP) server connects external AI coding assistants directly to your cloud infrastructure through a secure, hosted Streamable HTTP endpoint. Instead of copying logs and environment variables between your browser and editor, your AI agent can inspect deployments, diagnose build errors, stream container logs, manage secrets, and trigger production rollouts.
What URL do I connect my client to?
Connect your MCP client to https://mcp.deployxa.com/mcp. The server implements the Streamable HTTP transport with Server-Sent Events (SSE) and exposes discovery metadata at /.well-known/mcp.json.
How does authentication work?
The MCP server supports two authentication modes: 1) OAuth 2.1 with PKCE for interactive clients like Claude Desktop, Claude Code, and Cursor; 2) Deployxa API Keys (Bearer dxa_...) generated in your account dashboard under Settings -> API Keys.
Can I run the MCP server in read-only mode for safety?
Yes. If you want your AI assistant to only inspect projects, read logs, and check readiness scores without triggering deployments or changing secrets, generate a read-only API key or configure your client with DEPLOYXA_MCP_ENABLE_MUTATIONS=false.
Which AI clients are supported?
Any client implementing the Model Context Protocol specification works out of the box, including Claude Desktop, Cursor, Windsurf, Claude Code, Cline (VS Code), Gemini, and custom agents built on LangChain or CrewAI.
Can I also host my own custom MCP servers on Deployxa?
Yes! In addition to our official control plane MCP server, Deployxa is an ideal platform for deploying your own Streamable HTTP MCP server applications. Check our MCP Server framework deployment guide at /deploy/mcp-server.
Ready when you are
Give your AI coding assistants direct cloud control.
Connect your editor in 2 minutes. Inspect logs, diagnose build failures, and deploy apps autonomously.