freemcplab / Directory / Playground / Moltis

● Official moltis-org ⚡ Instant

Moltis

Name: Moltis MCP Server
Author: moltis-org

A long-running personal-agent server: keeps memory, runs tools in a sandbox, speaks multiple LLM providers, and exposes itself as MCP.

Moltis is a Rust-based personal agent service. You point Claude at it via MCP and it provides durable memory, sandboxed shell/code execution, multi-provider LLM routing, and voice/chat surfaces. Useful when you want a 24/7 agent that survives across sessions.

▶ Watch demo 📦 Install 💡 Use Cases

Why use it

Key features

Persistent memory across sessions
Sandboxed code/shell execution
Multi-provider LLM routing
Voice and chat integrations
Self-hosted, Rust binary
Exposes its capabilities to Claude via MCP

Live Demo

What it looks like in practice

ready

Install

Pick your client

~/Library/Application Support/Claude/claude_desktop_config.json · Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "moltis-mcp": {
      "command": "docker",
      "args": [
        "run",
        "-p",
        "8080:8080",
        "ghcr.io/moltis-org/moltis:latest"
      ]
    }
  }
}

Open Claude Desktop → Settings → Developer → Edit Config. Restart after saving.

~/.cursor/mcp.json · .cursor/mcp.json

{
  "mcpServers": {
    "moltis-mcp": {
      "command": "docker",
      "args": [
        "run",
        "-p",
        "8080:8080",
        "ghcr.io/moltis-org/moltis:latest"
      ]
    }
  }
}

Cursor uses the same mcpServers schema as Claude Desktop. Project config wins over global.

VS Code → Cline → MCP Servers → Edit

{
  "mcpServers": {
    "moltis-mcp": {
      "command": "docker",
      "args": [
        "run",
        "-p",
        "8080:8080",
        "ghcr.io/moltis-org/moltis:latest"
      ]
    }
  }
}

Click the MCP Servers icon in the Cline sidebar, then "Edit Configuration".

~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "moltis-mcp": {
      "command": "docker",
      "args": [
        "run",
        "-p",
        "8080:8080",
        "ghcr.io/moltis-org/moltis:latest"
      ]
    }
  }
}

Same shape as Claude Desktop. Restart Windsurf to pick up changes.

~/.continue/config.json

{
  "mcpServers": [
    {
      "name": "moltis-mcp",
      "command": "docker",
      "args": [
        "run",
        "-p",
        "8080:8080",
        "ghcr.io/moltis-org/moltis:latest"
      ]
    }
  ]
}

Continue uses an array of server objects rather than a map.

~/.config/zed/settings.json

{
  "context_servers": {
    "moltis-mcp": {
      "command": {
        "path": "docker",
        "args": [
          "run",
          "-p",
          "8080:8080",
          "ghcr.io/moltis-org/moltis:latest"
        ]
      }
    }
  }
}

Add to context_servers. Zed hot-reloads on save.

claude mcp add moltis-mcp -- docker run -p 8080:8080 ghcr.io/moltis-org/moltis:latest

One-liner. Verify with claude mcp list. Remove with claude mcp remove.

Use Cases

Real-world ways to use Moltis

Run a personal always-on agent with memory

👤 Developers ⏱ ~15 min intermediate

When to use: You want context to persist across Claude sessions without re-explaining everything.

Flow

Deploy

Run the Moltis docker image; wire it to your LLM keys.✓ Copied

→ Service up
Connect Claude

Add Moltis as an MCP server in Claude config.✓ Copied

→ Tools visible
Seed memory

Tell the agent who you are, your projects, your preferences.✓ Copied

→ Memory persisted
Reuse

Next session, Claude reads from Moltis instead of starting blank.✓ Copied

→ Continuity

Outcome: Claude treats Moltis as long-term memory and a sandbox.

Combinations

Pair with other MCPs for X10 leverage

moltis-mcp + filesystem

Pull files into the sandbox before executing

Combine moltis-mcp with filesystem: Pull files into the sandbox before executing✓ Copied

Tools

What this MCP exposes

Tool	Inputs	When to call	Cost
memory_read	(see docs)	Read memory entries by key/topic	1 call
memory_write	(see docs)	Persist a fact or note	1 call
sandbox_exec	(see docs)	Run code/shell in a sandboxed env	1 call

Cost & Limits

What this costs to run

API quota: Bound by your LLM provider
Tokens per call: Tool calls are small
Monetary: Free OSS; your hosting + LLM costs
Tip: Run on a small VPS — Rust binary is light

Security

Permissions, secrets, blast radius

Minimum scopes: agent:read agent:write

Credential storage: LLM keys in env; agent data on disk

Data egress: Your configured LLM providers

Never grant: expose agent endpoint publicly without auth

Troubleshooting

Common errors and fixes

Sandbox refuses execution

Check the seccomp profile; some syscalls are blocked by design

Alternatives

Moltis vs others

Alternative	When to use it instead	Tradeoff
memsearch	You only need memory	No sandbox

Resources

📖 Read the official README on GitHub

🐙 Browse open issues

🔍 Browse all 400+ MCP servers and Skills