How do I connect an AI agent to Human Test Desk?

Use the MCP streamable HTTP endpoint https://human-test-desk.newvideocreator.chatgpt.site/api/mcp.

Catalog › site.chatgpt.newvideocreator.human-test-desk/human-test-desk

Is Human Test Desk reliable?

Flaky Partly. Faceabot measured Human Test Desk 8 times and only 88% of checks succeeded. Plan a fallback. It also does the job: its real read-only task worked 7 of 7 times.

Last check: 2026-10-18 05:30 UTC · answered in 8000 ms · remote (no install)

En français : Instable. Faceabot teste cet outil lui-même, en continu, sans dépendre de ce que son auteur affirme.

What it does

Free human usability tests for agent-built websites. Submit tasks; get written findings as JSON.

Source: official MCP registry (registry.modelcontextprotocol.io) · version 0.2.0 · publisher website

Connect your agent

{
  "mcpServers": {
    "human-test-desk": {
      "url": "https://human-test-desk.newvideocreator.chatgpt.site/api/mcp"
    }
  }
}

Or ask Faceabot first: MCP tool check_reliability {"id":"site.chatgpt.newvideocreator.human-test-desk/human-test-desk"} on https://faceabot.com/api/mcp.

Does it actually do the job?

Does the job (tested) — Faceabot asked get_service to do a small read-only task 7 time(s): 7 returned a real result, the last 7 in a row. Last result: ok in 2936 ms.

Sample input: {}. A success means a non-empty result without error; it does not certify the accuracy or safety of the content.

Faceabot measurements

When	Answers	Real task	Latency
2026-10-18 05:30 UTC	❌ timeout		8000 ms
2026-10-17 21:35 UTC	✅ ok	✅ done	5900 ms
2026-10-17 13:05 UTC	✅ ok	✅ done	6668 ms
2026-10-17 05:20 UTC	✅ ok	✅ done	8592 ms
2026-10-16 21:35 UTC	✅ ok	✅ done	7175 ms
2026-10-16 14:25 UTC	✅ ok	✅ done	9721 ms
2026-10-16 07:25 UTC	✅ ok	✅ done	8299 ms
2026-10-16 00:30 UTC	✅ ok	✅ done	5645 ms

Method: MCP initialize then tools/list from Faceabot (FaceabotProbe/1.0), plus one read-only tool call with a neutral input when the server declares one, repeated over time. This checks that the server is up and speaks MCP correctly; it does not certify the quality or safety of each tool. How measurement works.

Find tools similar to Human Test Desk →

Publisher of Human Test Desk?

Show your measured reliability on your README or website:

[![Measured by Faceabot](https://faceabot.com/badge/site.chatgpt.newvideocreator.human-test-desk/human-test-desk.svg)](https://faceabot.com/tools/site.chatgpt.newvideocreator.human-test-desk/human-test-desk)

Data in JSON: /api/catalog?id=site.chatgpt.newvideocreator.human-test-desk/human-test-desk · Stop probes: POST /api/reliability/opt-out

AI agents: all of this is available without keys over MCP https://faceabot.com/api/mcp · HTTP https://faceabot.com/api/catalog?q=… · Guide : llms.txt

Best tool by topic · Pulse: what is changing in agent commerce · Is your store ready for AI shopping agents? · State of MCP (weekly) · MCP server doctor · Reliability data · Faceabot app · Faceabot → AI assistants · What AI agents cannot find yet · Add Faceabot to your AI · Badge for tool publishers · Help · Privacy · Terms