mcpField
M
Marsell

Marsell

Remote
docs.getmarsell.comBusiness & Marketing0Updated 2d ago

About Marsell

Marsell's hosted MCP server connects an AI client to Marsell projects for solo founders, small businesses, and teams. Depending on granted permissions, an AI can inspect project information and perform supported content, marketing, and sales actions.

Key features

  1. 01Read projects, brand, and market information
  2. 02Review analytics and pending decisions
  3. 03Permission-based project changes and actions
  4. 04Work with content, canvases, and videos
  5. 05Create images after a token quote

Use cases

  1. 01Ask what requires attention in a Marsell project
  2. 02Review brand and market context before planning work
  3. 03Check due homework before starting it
  4. 04Delegate permitted content or marketing actions

MCP support

The hosted Streamable HTTP MCP server at https://getmarsell.com/mcp exposes Marsell project tools to MCP clients. It supports OAuth or agent-key authentication, and available actions depend on permissions.

Configuration

Remote servers need no local install. Pick your client, then copy the config or install in one click.

Endpointhttps://getmarsell.com/mcp
Copy install command
Claude CodeTerminal
Cursor.cursor/mcp.json
VS Code.vscode/mcp.json
Claude Desktopclaude_desktop_config.json
Windsurf~/.codeium/windsurf/mcp_config.json

Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.

Tools

Details

From the project README

Marsell MCP

Marsell handles content, marketing and sales on autopilot for solo founders and small to medium businesses, with teams and organisations supported. Its hosted Model Context Protocol (MCP) server lets the AI you already use work with your Marsell projects.

This is Marsell's official public repository for connection instructions and MCP Registry metadata. The hosted backend is private and is not distributed in this repository.

Server URL: https://getmarsell.com/mcp
Transport: Streamable HTTP
Registry name: io.github.Marsell-Official/marsell

Full documentation · Try the interactive demo · Marsell

Connect

Sign in with OAuth

In an MCP client that supports remote servers and OAuth, add https://getmarsell.com/mcp, choose OAuth authentication, sign in to Marsell and review the permissions before pressing Allow. Do not select No authentication. Client plans and administrator permissions can affect connector availability; see the connection guide.

For Claude Code:

claude mcp add --transport http marsell https://getmarsell.com/mcp

Then run /mcp, select marsell, and choose Authenticate.

Connect with an agent key

In Marsell, open a project, go to Agents → Your own AI, and choose Make a key. Choose Reads only, or Can act if your agent should also change things and spend tokens. Copy the key when it is shown; it is only displayed once.

For Codex, set the environment variable in the shell that starts Codex:

export MARSELL_KEY='YOUR_MARSELL_AGENT_KEY'
codex mcp add marsell --url https://getmarsell.com/mcp --bearer-token-env-var MARSELL_KEY

On PowerShell, set the variable with $env:MARSELL_KEY = 'YOUR_MARSELL_AGENT_KEY', then run the same codex mcp add command.

Alternatively, configure Codex in ~/.codex/config.toml:

[mcp_servers.marsell]
url = "https://getmarsell.com/mcp"
bearer_token_env_var = "MARSELL_KEY"

Other clients can use the server URL with this HTTP header:

Authorization: Bearer YOUR_MARSELL_AGENT_KEY

Replace the placeholder with your own key through your client's secure credential settings. Never commit keys to Git or paste them into an issue.

What your AI can do

  • Read projects, brand information, market research, people, analytics, token balances and decisions waiting for you.
  • With permission, answer decisions, update brand information, plan work and use Marsell's content, marketing and sales tools.
  • Create pictures with a token quote first, and work with Marsell's chat on content, canvases and videos.
  • Set up a business using the context and files your AI already has.
  • Pick up homework that needs your own AI's logins or files, and report its results to Marsell.

The tools available depend on your permissions. Your client can discover the current list through MCP tools/list; the full documentation describes the tools and their approval requirements.

Try asking:

What needs me on Marsell this week?

Read my brand and market, then tell me what we should work on next.

Check my Marsell homework and show me what is due before starting.

Homework does not create a background schedule by itself. If you want recurring checks, your AI client must support scheduling; it can report that schedule to Marsell with homework_schedule.

Permissions and account access

Reads only lets your AI inspect information. Can act also allows supported changes and actions that can spend tokens. Tool-specific approval requirements still apply, and Marsell checks your current account role on every call. Team and organisation access follows your Marsell permissions.

You can disconnect an OAuth app or revoke a key under Agents → Your own AI → Connected apps and keys. Account, plan and token requirements apply to the hosted service.

Troubleshooting

  • 401 Unauthorized: authenticate with OAuth or supply a valid agent key. The endpoint requires authentication.
  • Opening the URL in a browser fails: this is a Streamable HTTP MCP endpoint, not a web page. Use an MCP client.
  • A tool is missing or an action is denied: check the connection's scope, your current role and the tool's requirements.
  • A key stopped working: check whether it was revoked and whether the client can still read its environment variable.

OAuth discovery is available at protected resource metadata and authorization server metadata.

Repository and registry maintenance

server.json describes the hosted server for the official MCP Registry. It does not install or expose Marsell's backend.

Connection instructions and registry metadata are included in Marsell's MCP staleness checks. When the hosted interface changes, maintainers review this README and the manifest against the implementation. Registry metadata changes require validation with the official mcp-publisher and a new version before publication; changes in this GitHub repository alone do not update the registry.

Please report connection or documentation issues through this repository's issues, without credentials or private project data.

License

The documentation and metadata in this repository are licensed under MIT. That license does not cover the hosted Marsell service, its private backend, or Marsell's trademarks and brand assets.

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.