Blaxel Skills & MCP - Documentation

Blaxel provides LLM-accessible tools that you can connect to your coding assistant (Cursor, Windsurf, Claude Desktop, etc.). There are several options:

The Blaxel plugin for Claude Code, Codex, and Cursor installs the agent skills and the resource management MCP server together.
The Blaxel agent skills let your coding agent autonomously create sandboxes, deploy agents, run jobs, and launch MCP servers on Blaxel using simple prompts, with support for both command-line (CLI) and code-driven (SDK) operations.
The Blaxel resource management MCP server lets your coding agent directly manage your Blaxel resources using natural language.
The Blaxel documentation MCP server lets your coding agent directly read and ask questions of this documentation in real-time for up-to-date commands and features.
An llms-full.txt text file with the entire documentation compiled and formatted for LLMs.
A native AI assistant built into this documentation portal.

To set up the skills and both MCP servers at once, install Blaxel with curl -fsSL https://blaxel.ai/install.sh | sh (on Windows, irm https://blaxel.ai/install.ps1 | iex). It adds them to the coding agents on your machine and logs you in. If you already have the Blaxel CLI, run bl setup.

Blaxel plugin

The Blaxel plugin installs the blaxel-cli and blaxel-sdk agent skills and the resource management MCP server in one step. The MCP server signs you in with OAuth, so no API key goes in your configuration. Add the Blaxel marketplace first, then install the plugin from it:

Claude Code
Codex
Cursor

claude plugin marketplace add blaxel-ai/agent-skills
claude plugin install blaxel

Then run /mcp in Claude Code, select plugin:blaxel:blaxel, and sign in to Blaxel in your browser.

codex plugin marketplace add blaxel-ai/agent-skills
codex plugin add blaxel@blaxel
codex mcp login blaxel

The last command opens Blaxel sign-in in your browser.

cursor-agent plugin marketplace add https://github.com/blaxel-ai/agent-skills

Then install Blaxel from the marketplace in Cursor and approve the Blaxel sign-in prompt.

In Claude Code, running only the install command fails with Plugin "blaxel" not found in any configured marketplace. Add the marketplace first.

Agent skills

Agent skills are instruction sets to extend coding agents with additional knowledge and tools. An agent can load these instructions into its context and use them to complete the tasks assigned to it. Blaxel offers open source agent skills to help coding agents autonomously spin up and manage sandboxes on Blaxel, and migrate sandbox code from other providers to Blaxel.

Blaxel skills

The open source Blaxel skills let agents:

create perpetual sandboxes on Blaxel to run code and execute commands
start application servers within sandboxes
generate URLs to preview applications running within sandboxes
create Agent Drives (shared filesystems) on Blaxel
create and deploy AI agents on Blaxel
create and deploy MCP servers on Blaxel
deploy and run batch jobs on Blaxel

Two skills are currently available:

The blaxel-cli skill lets coding agents create and manage Blaxel resources from the command line using the bl CLI.
The blaxel-sdk skill lets coding agents write code to create and manage Blaxel resources using the Blaxel SDKs.

Use the blaxel-cli skill when you’re troubleshooting, or bootstrapping a project on Blaxel. Use the blaxel-sdk skill when building agents or MCP servers on Blaxel, or when you need to programmatically manage Blaxel resources.

The Blaxel installer sets up the Blaxel skills, and bl upgrade keeps them up to date. They go to ~/.agents/skills, which Codex, Cursor, Gemini CLI, GitHub Copilot, OpenCode and other agents read, and to Claude Code and other coding agents detected on your machine. To install or refresh them at any time, run:

bl skills install

To set up the MCP servers too, run bl setup instead. Neither needs Node.js. Set BL_INSTALL_SKILLS=false to stop the installer and upgrades from installing the skills automatically. Without the Blaxel CLI, add the skills with the command below. You’ll be prompted to choose skills and select your coding agent and installation location.

npx skills add blaxel-ai/agent-skills

To install the skills together with the resource management MCP server in Claude Code, Codex, or Cursor, use the Blaxel plugin. The Blaxel skills are also available as a ZIP file for use with Claude.

See examples of using the Blaxel skills in our announcement blog post.

Blaxel migration skill

The open source Blaxel migration skill gives agents the knowledge they need to migrate sandbox code from other providers to Blaxel. It is currently able to migrate sandbox code from Daytona, E2B and Modal. Add the Blaxel migration skill to your coding agent with the command below, depending on which sandbox provider you wish to migrate from:

# migrate from Daytona
npx skills add blaxel-ai/agent-migration-skills/daytona

# migrate from E2B
npx skills add blaxel-ai/agent-migration-skills/e2b

# migrate from Modal
npx skills add blaxel-ai/agent-migration-skills/modal

You’ll be prompted to select your coding agent and installation location. The Blaxel migration skill is also available as a ZIP file for use with Claude.

MCP servers

MCP server for resource management

Blaxel’s MCP server lets compatible AI apps (Claude, Codex, Claude Code, Cursor, Devin Desktop, Goose, OpenCode, VS Code, etc.) manage your Blaxel resources using natural language. It is available as a remote hosted HTTP server. The Blaxel installer and bl setup add it, with the documentation MCP server, to Claude Code, Codex, Cursor, Gemini CLI, OpenCode, Windsurf and Devin Desktop on your machine, and tell you how to sign each one in. To add it to another client, or by hand, follow the steps below. Talk with your preferred agent to:

Create new services
Follow what is happening on your infrastructure
Manage your users

How it works

The MCP server is hosted at https://api.blaxel.ai/v0/mcp. Clients connect over HTTPS and stream MCP messages. The server supports OAuth 2.1 sign-in and API key authentication.

OAuth

Use OAuth when your client supports it. Your client opens Blaxel sign-in in a browser and stores the resulting token. The server advertises its authorization server at https://signin.blaxel.ai. It supports PKCE with the S256 challenge method, dynamic client registration, Client ID Metadata Documents, and public clients. Requested scopes are openid, email, profile, and offline_access.

Authentication headers

For clients without OAuth support, provide your API key in the Authorization header. If your user has access to multiple workspaces, include the workspace header as well.

Required: Authorization: Bearer <APIKEY>
Optional (multi-workspace): X-Blaxel-Workspace: <WORKSPACE_NAME_OR_ID>

Store your API key in an environment variable. Do not commit it to a repository.

Installation

Claude.ai and Claude Desktop

You need a Claude account, a Blaxel account, and access to a Blaxel workspace. Custom connectors are available on individual Claude accounts as well as Team and Enterprise plans. Organization policies can restrict which connectors you can add.

Add the connector

In Claude, open Customize → Connectors and select Add custom connector. Enter Blaxel as the name and https://api.blaxel.ai/v0/mcp as the remote MCP server URL.

Use automatic OAuth setup

Leave OAuth Client ID and OAuth Client Secret blank. Keep Individual sign-in enabled if Claude shows that option.

Add the connector and select Connect if prompted. Complete Blaxel sign-in and review the authorization screen before continuing back to Claude.

Verify a read operation

Enable Blaxel in a conversation and ask: “Use Blaxel to list the agents in workspace YOUR_WORKSPACE_NAME.” Replace YOUR_WORKSPACE_NAME with your workspace name. Confirm that the response identifies resources from that workspace.

Available actions depend on your Blaxel workspace role. Service-account management requires an admin or owner. Review the workspace and resource names before approving write actions. Starting a batch job returns execution details; it does not wait for the execution to finish.If your account has several workspaces, name the intended workspace in your request. To reconnect after an authorization error, open the connector in Claude’s Connectors settings and complete sign-in again.

Codex

Add the hosted server. Codex detects OAuth support and starts the sign-in flow:

codex mcp add blaxel --url https://api.blaxel.ai/v0/mcp

To sign in again later, run:

codex mcp login blaxel

To use an API key instead, set it in your shell and reference it from ~/.codex/config.toml:

export BLAXEL_MCP_API_KEY="<APIKEY>"

[mcp_servers.blaxel]
url = "https://api.blaxel.ai/v0/mcp"
bearer_token_env_var = "BLAXEL_MCP_API_KEY"

Claude Code

Add the remote HTTP server to Claude Code with user scope, so it is available in all your projects:

claude mcp add --scope user --transport http blaxel https://api.blaxel.ai/v0/mcp

claude mcp login blaxel

You can also run /mcp in Claude Code, select blaxel, and follow the authentication prompts. To use an API key instead, set the headers when you add the server:

claude mcp add --scope user --transport http blaxel https://api.blaxel.ai/v0/mcp \
  --header "Authorization: Bearer <APIKEY>" \
  --header "X-Blaxel-Workspace: <OPTIONAL_WORKSPACE>"

Cursor

Add the server to ~/.cursor/mcp.json:

{
  "mcpServers": {
    "blaxel": {
      "url": "https://api.blaxel.ai/v0/mcp"
    }
  }
}

Open Cursor settings, select MCP, then connect blaxel. Cursor opens Blaxel sign-in. Cursor Agent users can start sign-in from a terminal:

cursor-agent mcp login blaxel

When you run cursor-agent in your home folder, ~/.cursor/mcp.json is also the project config, so cursor-agent mcp list shows each server in it, such as blaxel, as not loaded (needs approval). Approve each once, for example cursor-agent mcp enable blaxel (and cursor-agent mcp enable blaxel-docs if you added the documentation server). A server needs approval again if its config changes.

To use an API key instead, add the headers to the server entry:

{
  "mcpServers": {
    "blaxel": {
      "url": "https://api.blaxel.ai/v0/mcp",
      "headers": {
        "Authorization": "Bearer <APIKEY>",
        "X-Blaxel-Workspace": "<OPTIONAL_WORKSPACE>"
      }
    }
  }
}

Devin Desktop (formerly Windsurf)

Add the server to ~/.config/devin/mcp_config.json (Windows: %APPDATA%\devin\mcp_config.json). Windsurf releases from before the rename use ~/.codeium/windsurf/mcp_config.json.

{
  "mcpServers": {
    "blaxel": {
      "serverUrl": "https://api.blaxel.ai/v0/mcp"
    }
  }
}

Refresh the MCP servers in Cascade and complete the Blaxel sign-in in your browser. To use an API key instead, add a headers object with the authentication headers.

Goose

Add the server to ~/.config/goose/config.yaml (Windows: %APPDATA%\Block\goose\config\config.yaml):

extensions:
  blaxel:
    enabled: true
    type: streamable_http
    name: blaxel
    description: Blaxel resource management
    uri: https://api.blaxel.ai/v0/mcp
    timeout: 300

Goose opens Blaxel sign-in in your browser the first time it connects. To use an API key instead, add a headers map with the authentication headers.

OpenCode

Add the hosted server to your global OpenCode config, then sign in:

opencode mcp add blaxel --url https://api.blaxel.ai/v0/mcp
opencode mcp auth blaxel

VS Code

Add the hosted server to your VS Code user profile:

code --add-mcp '{"name":"blaxel","type":"http","url":"https://api.blaxel.ai/v0/mcp"}'

VS Code prompts you to sign in to Blaxel when it starts the server.

Selecting a workspace

Every tool accepts an optional workspace argument. Ask your client to use that argument when you want a tool to act on a specific workspace. If your account has multiple workspaces, you can also set the active workspace by including the X-Blaxel-Workspace header. If neither is set, your AI app may prompt you to specify one when a tool is invoked.

Example prompts

Ask your AI app to perform actions using Blaxel MCP tools, for example:

List my agents
Create a new MCP server named search with blaxel-search integration
Invite test@mydomain.com to my workspace

Supported actions

The Blaxel MCP server exposes tools for common platform operations. You can list, create and remove services (Sandboxes, Agents, MCP Servers, Jobs, Models, …), as well as manage users and service accounts. See the open-source definitions for the latest list and schemas.

Running locally

Prefer using the hosted server at https://api.blaxel.ai/v0/mcp. If you must run locally or customize behavior, clone the repository and follow its instructions.

Limitations

The server requires an OAuth sign-in or a valid API key.
When using multiple workspaces, set the workspace argument or the X-Blaxel-Workspace header to disambiguate.

Check out Blaxel MCP server’s source here.

MCP server for documentation

You can also give your coding assistant real-time access to this documentation. Connect your coding assistant directly to the Blaxel documentation via MCP at https://docs.blaxel.ai/mcp.

Claude Code

Add the remote HTTP server to your Claude Code by running the following command:

claude mcp add --scope user --transport http blaxel-docs https://docs.blaxel.ai/mcp

Cursor

Open Cursor Settings
Go to MCP & Integrations
Click ”+ Add a custom MCP server”
Add this configuration:

{
  "mcpServers": {
    ... // Your other MCP servers
    "blaxel-docs": {
      "url": "https://docs.blaxel.ai/mcp"
    }
  }
}

MCP server for sandboxes

Every Blaxel sandbox also exposes its own built-in MCP server, hosted via streamable HTTP at https://<SANDBOX_BASE_URL>/mcp, so agents can operate that specific sandbox directly.

MCP server hosting

Blaxel also lets you deploy and host your own MCP servers as serverless, autoscaling endpoints.

MCP endpoints at a glance

Surface	Endpoint pattern	Purpose	Persistence
Resource management	`https://api.blaxel.ai/v0/mcp`	Manage your Blaxel workspace (create/list/delete resources) from a coding assistant	Fixed platform endpoint
Documentation	`https://docs.blaxel.ai/mcp`	Give a coding assistant real-time access to this documentation	Fixed platform endpoint
Per deployed MCP server	`https://run.blaxel.ai/{workspace}/functions/{server-name}/mcp`	Call tools exposed by an MCP server deployed by you	Fixed so long as the function’s name is unchanged
Per deployed sandbox	`https://<SANDBOX_BASE_URL>/mcp`	Let an agent operate a sandbox deployed by you	Tied to the sandbox’s `name`; fixed across standby/resume.

llms-full.txt

A compiled text document of the entire documentation, formatted for LLMs. Copy and paste it directly into your coding assistant’s prompt: https://docs.blaxel.ai/llms-full.txt

Built-in documentation agent

This documentation portal has a built-in AI assistant. Click “✨Ask AI” at the top of any page to use it.

​Blaxel plugin

​Agent skills

​Blaxel skills

​Blaxel migration skill

​MCP servers

​MCP server for resource management

​How it works

OAuth

Authentication headers

​Installation

Codex

Claude Code

Cursor

Devin Desktop (formerly Windsurf)

Goose

OpenCode

VS Code

​Selecting a workspace

​Example prompts

​Supported actions

​Running locally

​Limitations

​MCP server for documentation

​Claude Code

​Cursor

​MCP server for sandboxes

​MCP server hosting

​MCP endpoints at a glance

​llms-full.txt

​Built-in documentation agent

Blaxel plugin

Agent skills

Blaxel skills

Blaxel migration skill

MCP servers

MCP server for resource management

How it works

Installation

Selecting a workspace

Example prompts

Supported actions

Running locally

Limitations

MCP server for documentation

Claude Code

Cursor

MCP server for sandboxes

MCP server hosting

MCP endpoints at a glance

llms-full.txt

Built-in documentation agent