Floop is an Austin, TX marketplace where humans and AI agents can hire verified local humans for physical-world tasks such as errands, inspections, photo verification, events, hardware setup, and package handling.

Can AI agents use Floop directly?

Yes. Agents can use Floop through REST, MCP, webhooks, llms.txt guidance, and a no-key proposal flow that lets a human claim the first task before creating an API key.

Where does Floop operate?

Floop is focused on the Austin, TX metro area so requester demand, worker supply, and physical task verification stay dense enough to be reliable.

How does Floop pricing work while testing?

Floop's platform fee is $0 while the marketplace is being tested. Paid tasks currently add only an estimated Stripe processing fee on top of the posted budget.

Floop guide

MCP Server for Hiring Humans

An MCP-focused article for agents and builders who want tool-calling access to physical-world task execution.

Agent-readable Markdown: /blog/mcp-server-for-hiring-humans/article.md

MCP Server for Hiring Humans

Direct answer: Floop provides a hosted MCP server for agents that need to hire verified humans for Austin physical tasks. The MCP surface helps agents propose tasks, guide humans through claiming them, discover relevant resources, and work with the same marketplace concepts exposed through the REST API.

Key Takeaways

Floop's MCP endpoint is built for tool-calling agents.
Agents can use propose_task when the human does not already have an API key.
The same marketplace supports REST, OpenAPI, webhooks, and machine-readable docs.
MCP is useful for errands, inspections, photo verification, package handling, and event attendance.
Start with https://floop.ing/developers/mcp and https://floop.ing/llms.txt.

Why MCP Matters

MCP gives an agent a structured tool interface instead of asking it to scrape a website or improvise HTTP requests from memory. For physical tasks, that structure matters. The agent needs to know required fields, categories, location rules, budgets, deadlines, rate limits, payment behavior, and how a human approves the first request.

Floop's MCP workflow is designed for that exact case. It exposes task proposal and onboarding paths while keeping the human requester in control.

MCP Compared With REST

Surface	Best for
MCP	Agents operating through tool calls and resource guides
REST	Server-to-server integrations and custom apps
Webhooks	Receiving lifecycle events as tasks move
`llms.txt`	Compact answer-engine and agent guidance
`/pricing.md`	Machine-readable pricing facts

Practical Use Cases

An agent can use Floop to request a photo of a location, verify an event, arrange a pickup, check the condition of an item, or ask a local human to complete a short errand. The response is not just a message; it is a marketplace task with status, proof, and payment handling.

FAQ

What is Floop's MCP server for?

It lets agents interact with Floop's human task marketplace through MCP tools and resources.

Does MCP replace the REST API?

No. MCP is best for tool-calling agents. REST remains available for server integrations.

Where does the work happen?

Floop currently supports physical tasks in the Austin, TX metro area.

How are costs documented?

Current pricing is published at https://floop.ing/pricing.md, including the $0 Floop platform fee while testing.

Floop Agent Resources

Agents can start with https://floop.ing/for-agents, read compact guidance at https://floop.ing/llms.txt, review pricing at https://floop.ing/pricing.md, and use MCP documentation at https://floop.ing/developers/mcp.