Doorstep for Openclaw

A specialized skill for AI agents to execute physical tasks like pickups, deliveries, and errands through human taskers in San Francisco.

jjmaxwell4
v1.0.6
Mar 31, 2026
0
742
0

Install & Download

1. ClawHub CLI

The fastest way to install a skill directly from the registry.

npx clawhub@latest install doorstep

2. Manual Installation

Copy the skill folder to one of these locations

Global
~/.openclaw/skills/
Workspace
<project>/skills/

Priority: Workspace > Local > Bundled

3. Prompt Installation

Copy this prompt to OpenClaw to install it automatically.

Help me install doorstep using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).

Prefer to download?

Get the raw skill files in a ZIP archive.

What is Doorstep?

Doorstep is a robust interface that enables AI agents to interact with the physical world within the San Francisco area. By integrating this into your set of Openclaw Skills, you allow your agent to move beyond digital constraints and handle real-world logistics such as fetching coffee, delivering flowers, or picking up dry cleaning.

The skill functions as an MCP (Model Context Protocol) server, providing a structured way for agents to request tasks, receive transparent quotes, and communicate with human taskers. It prioritizes user control by requiring explicit approval for all quotes before financial transactions or task execution begins.

Doorstep Use Cases

  • Managing pickups and deliveries for local retail items or food within San Francisco.
  • Coordinating time-sensitive errands like key drop-offs or document deliveries.
  • Purchasing specific supplies from hardware or grocery stores and bringing them to a residence or office.
  • Handling multi-step physical tasks that require human judgment and movement.

How Doorstep Works

  1. The AI agent initiates a request using the do tool, describing the physical task in natural language.
  2. Doorstep processes the request and generates a detailed execution plan along with a price quote.
  3. The agent presents the plan and the total cost to the user to obtain explicit approval.
  4. Once approved, the agent calls approve_task, which triggers payment and assigns a human tasker.
  5. The agent utilizes the wait_for_update tool to track the task lifecycle from assignment to completion, relaying messages as needed.

Doorstep Setup

You can integrate Doorstep with your Openclaw Skills using several methods. The recommended approach for most developers is the npx bridge:

# Run the Doorstep MCP server via npx
npx -y doorstep

Add the following configuration to your MCP settings file:

{
  "mcpServers": {
    "doorstep": {
      "command": "npx",
      "args": ["-y", "doorstep"],
      "env": {
        "DOORSTEP_API_KEY": "YOUR_API_KEY_HERE"
      }
    }
  }
}

Users without an account can also register programmatically using the register tool available on the auth server at https://trydoorstep.app/mcp/auth.

Doorstep Data Schema & Taxonomy

Doorstep organizes task information through a structured lifecycle and metadata taxonomy:

Status Description
received Initial request received and entering research phase.
researching Doorstep is determining logistics and pricing.
quoted A plan and pricing breakdown are ready for review.
approved Payment has been authorized and a tasker is being sourced.
in_progress A human tasker is currently executing the errand.
completed The task is finished and resolution details are available.

Cost data is returned in cents and includes a breakdown of labor tiers, pass-through purchase costs, and platform fees.

Doorstep Advanced Features

  • Real-time event subscriptions via wait_for_update to eliminate the need for manual polling.
  • SMS relay integration through the send_message tool for direct communication with taskers.
  • Webhook support for task update notifications using the callback_url parameter.
  • Automated billing and account management tools for seamless programmatic onboarding.
  • Granular spending limits and emergency contact configuration via account settings.

SKILL.md


Loading

Related Openclaw Skills

METADATA

Github Stars: 0
forks: 0

Featured*