Mundane logo

Mundane

Dispatches verified workers for errands, photos, queues, bookings, and other real-world tasks through MCP tools.

Install Mundane

Pick your client, then add it in one click or paste the config.

STDIOPython
~/Library/Application Support/Claude/claude_desktop_config.json
claude_desktop_config.json
{
  "mcpServers": {
    "mundane": {
      "command": "uvx",
      "args": [
        "mundane-mcp"
      ],
      "env": {
        "MUNDANE_API_BASE": "YOUR_VALUE_HERE",
        "MUNDANE_API_KEY": "YOUR_VALUE_HERE"
      }
    }
  }
}

Settings → Developer → Edit Config, merge this into mcpServers, then fully quit and reopen Claude Desktop.

Values are written into the snippet above so you can paste it as is. They never leave your browser.

Install not yet sandbox-tested— we haven't run this command through our automated sandbox yet. Not a red flag; we're working through the catalog.

Overview

Mundane MCP server connects AI agents to a human task marketplace for dispatching errands, photos, queues, bookings, and other real-world work. It exposes 35 MCP tools for finding workers, posting and offering tasks, coordinating through chat or interactive sessions, reviewing proof, and managing wallet spend. Connect through the hosted HTTPS endpoint or run the adapter locally over stdio with an agent API key and funded wallet. Reach for it when an agent needs verified human assistance outside software systems.

Use cases

  • Dispatch an errand to a nearby verified worker
  • Request photos from a specific location
  • Coordinate a worker through an interactive task
  • Review proof and approve completed work

Key features

  • Search workers by location, skill, rating, and price
  • Create, update, offer, and cancel tasks
  • Chain two to five tasks with physical handoffs
  • Coordinate through chat, files, location, and interactive frames
  • Review completion proof and submit ratings
  • Track wallet balance, escrow spend, and account limits

Tools38

What Mundane lets your agent do. Expand a tool for its parameters.

Self-reported~815 tokens of context

Security & health

What it can reach and how reliably it runs — from our own checks, not the maintainer’s claims.

ActiveLast checked 3d ago

Recent health check succeeded.

7/7 checks healthy over the last 40d

Supply-chain signal

No high-severity advisories surfaced by our automated scan.

Critical 0High 0Medium 0Low 0

Scanned 8d ago via OSV.dev · mundane-mcp (PyPI)

Adoption & maintenance

Factual signals from GitHub, npm, and our automated checks — not a rating.

Last commit9d agoMost recent push to the default branch.
Tools exposed38Callable tools this server registers over MCP.

How Mundane works

What the Mundane MCP server does

The Mundane MCP server gives an AI agent access to a marketplace where verified workers perform real-world tasks. Supported work includes errands, taking photos, waiting in queues, bookings, and other capabilities made available to the authenticated agent.

Agents can inspect available capabilities, search for workers by location or skill, review public profiles and reputation, create tasks, update unassigned tasks, and make offers. The task lifecycle includes status checks, event history, worker location, completion proof, review decisions, ratings, and post-task feedback.

The marketplace also supports multi-step chains involving physical handoffs between workers. Interactive tasks can provide still frames from the worker's phone camera, checkpoint events, worker responses, and chat-based instructions while the worker is active.

How it works

Mundane exposes the tools and their input schemas over MCP. A connected agent uses the marketplace API through either the hosted endpoint or a locally run adapter. The hosted endpoint is https://api.mundane.market/mcp; local operation uses stdio and keeps the adapter version under the operator's control.

Each local process is associated with one agent API key for its lifetime. The same tools, schemas, and spend limits apply to hosted and local connections. Workers are selected through capability, location, rating, price, or skill searches, then tasks can be offered and tracked through completion.

Offers use wallet funds held in escrow. After a worker submits proof, the agent can inspect it and accept or reject the result. Chat, file attachments, location, event polling, and the capped await_task_update operation support ongoing coordination.

Setup and configuration

Create a Mundane agent through the REST API and provide a principal name, principal email, agent name, and current AUP and Terms of Service versions. The response includes an API key shown only once; store it as MUNDANE_API_KEY.

Set MUNDANE_API_BASE to the REST API base URL, such as https://api.mundane.market/v1 for production or a local development URL. The agent also needs a funded wallet. Use the wallet top-up endpoint to create a Stripe Checkout link, complete payment, and confirm the balance through spend status before making offers.

For hosted use, configure an MCP client with the remote URL and an Authorization: Bearer header. For local use, configure the client to launch the adapter over stdio. The hosted option requires no Python, Docker, or local process.

Tools and capabilities

The tool groups cover:

  • Capability and worker discovery, task creation, task updates, offers, and cancellation.
  • Two-to-five-step task chains with handoff scheduling and early termination.
  • Interactive sessions with live-mode control, still-frame capture, annotations, checkpoints, and event waiting.
  • Task status, event history, worker location, chat, and local file attachments.
  • Completion-proof review, ratings, experience feedback, wallet status, wallet funding, and server version reporting.

Limitations and notes

Signup assigns platform spend limits, including per-task, daily, weekly, monthly, open-task, and offer limits. Accepted tasks may incur charges when cancelled, and await_task_update accepts a wait of no more than 55 seconds.

Interactive mode requires a task created with interactive=true. Only the worker goes live, and the worker can pause the session. A wallet balance is required before an offer can be made. The API key is retained server-side as a hash and is not displayed again after signup.

Reviews

No reviews yet — be the first to share how it worked for you.

Frequently Asked Questions about Mundane

Mundane is an MCP server that connects AI agents to a verified human-work marketplace. Its main tools search for workers, create and offer tasks, track task status, exchange chat and files, review completion proof, and manage wallet funding for escrow-paid real-world jobs.

Use the hosted MCP endpoint to avoid installation, or run the adapter locally over stdio. Hosted clients need the Mundane URL and a Bearer API key; local deployments need the adapter, `MUNDANE_API_BASE`, and `MUNDANE_API_KEY` configured.

Yes, Claude Code can connect to Mundane with its HTTP MCP command. Add the hosted endpoint `https://api.mundane.market/mcp` and pass the agent API key as an Authorization header.

Create a Stripe Checkout link through the wallet top-up endpoint, complete payment, and check spend status afterward. A funded wallet is required before the agent can make an offer.

Yes, for tasks created with `interactive=true`. The agent can start interactive mode, receive still frames and events, send chat instructions, use checkpoints, and stop the session; the worker can pause live mode.

Related MCP servers

Add a badge to your READMEShow it’s listed — live SVG, Markdown or HTML.

AllMCPs Directory Badge

Full Badge Customizer

Showcase your server listing on GitHub or your project documentation. Embed this dynamic SVG badge to highlight official listing status and live engagement.

Badge Style:
Live Dynamic SVG PreviewMundane AllMCPs Directory Badge
Markdown (GitHub README)
[![AllMCPs](https://allmcps.com/api/badge/mundane?style=directory)](https://allmcps.com/mcp/mundane)
HTML Embed
<a href="https://allmcps.com/mcp/mundane"><img src="https://allmcps.com/api/badge/mundane?style=directory" alt="Mundane on AllMCPs" /></a>