Twothumbs

io.github.asselwanPublished by GitHub account asselwanv1.0.0Updated Oct 25, 2026

Verifies AI agent work end to end: real artifacts and outcomes checked, not self-reported success.

VerifiedStreamable HTTPWeb executableBrowser AutomationDeveloper Tools

Overview

AI-generated overview

A hosted MCP server that verifies what an AI agent actually did on a public web surface, returning a pass/fail verdict based on real outcomes rather than self-

What you can do
Run a full-cycle web verdict on a public URL, exercising visible affordances such as clicking and submitting forms with synthetic data, then judging the real downstream outcome instead of a self-reported success message. List the target surfaces currently configured for verdict runs. Redeem a promotional access code for an API key. With an API key, additional tools for retrieving the last verdict, checking balance and setting auto top-up are unlocked.
When to use it
Useful when you want an independent check that an agent's web-facing work actually produced the intended result, for example after form submissions or multi-step interactions on a public site, rather than trusting the agent's own success claim.
Requirements
A remote Streamable HTTP endpoint; no local install, package or API key is needed for the anonymous tools. An API key, obtained by redeeming an access code or requesting one, unlocks extra tools. Network access to the hosted service.
Before you install
The verdict tool actively interacts with public web surfaces, clicking and submitting forms with synthetic data, so it can trigger real actions on those sites. Paid credit packs and auto top-up involve spending money. An API key is a credential to keep private.

Installation

In SourceWeft

  1. Open Twothumbs in the dashboard and add it to a workspace.
  2. Enable the server for the chats that should use its tools.

Web executable via Streamable HTTP. Remote servers run from the web runtime once configured in a workspace.

Use with an agent

Paste this prompt into Claude Code, Codex, Cursor or another agent — or into SourceWeft chat. The agent reads this server's install guide, shows you its publisher, endpoint and sign-in, and adds it to your client once you agree. Keys never go into a command.

Read https://sourceweft.com/mcp/io.github.asselwan%2Ftwothumbs/install.md and connect the MCP server it describes to my client. Before changing any config, show me its publisher, endpoint and sign-in method and wait for my OK. Never put an API key in a command or in a file I would commit; ask me where it should come from.

Read the install guide the agent follows

Other MCP clients

Add this to your client's mcpServers config.

{
  "mcpServers": {
    "twothumbs": {
      "type": "http",
      "url": "https://twothumbs.co/mcp"
    }
  }
}

README

TwoThumbs MCP

Verify what your AI agent actually did — not what it says it did.

TwoThumbs is a hosted, remote MCP server built by NOMOI Labs. It engineers an evidence-backed prompt, challenges weak premises, then checks the real artifact and downstream outcome a public web surface actually produced — instead of trusting a self-reported "success" message. It also carries a much larger internal prompt-engineering and verification-lifecycle tool set (51 tools total), of which the 3 tools below are exposed publicly over the hosted HTTP endpoint.

This repository is a thin discovery wrapper: an MCP client config, a Claude Code plugin manifest, and this README. There is no source code to install or build — the server itself is hosted and run by NOMOI at twothumbs.co.

Connect

The server speaks Streamable HTTP at:

https://twothumbs.co/mcp

No local install, no package, no API key required to use the anonymous tools.

Claude Code

bash
claude mcp add --transport http twothumbs https://twothumbs.co/mcp

Or install this repository as a Claude Code plugin (adds the same server via .mcp.json):

/plugin marketplace add asselwan/twothumbs-mcp/plugin install twothumbs@twothumbs-mcp

Cursor

Add to your MCP settings (Cursor Settings → MCP):

json
{  "mcpServers": {    "twothumbs": {      "url": "https://twothumbs.co/mcp"    }  }}

Generic MCP client

json
{  "mcpServers": {    "twothumbs": {      "type": "http",      "url": "https://twothumbs.co/mcp"    }  }}

Works with any client that supports the Streamable HTTP transport — Claude Code, Cursor, Codex, and CI/headless clients included.

Tools

Available anonymously, with a free daily quota:

ToolWhat it does
run_verdictExercises every visible affordance on a public URL — clicking, submitting forms with clearly marked synthetic data — then returns a pass/fail verdict grounded in the real downstream outcome, not a self-reported success message.
list_surfacesLists the target surfaces currently configured for run_verdict.
redeem_codeExchanges an access code for an API key.

An API key additionally unlocks get_last_verdict, get_balance, and set_auto_topup.

Pricing

  • Free — 3 surface scans/day, no signup required.
  • $9 — 100 credits.
  • $29 — 500 credits.

Billing is invoice-based; email through twothumbs.co to request a key, or use redeem_code with an access code.

Links

License

The contents of this repository (plugin manifest, client config, documentation) are MIT licensed — see LICENSE. This license covers only this wrapper repository; it does not extend to the TwoThumbs product or hosted service, which remain the property of NOMOI.

Source: README.md at commit f722748

Tools

3

run_verdict

Run a full-cycle web verdict

Parametersurlsurfaceregistermax_pages

list_surfaces

List configured surfaces and target URLs

Takes no parameters.

redeem_code

Redeem a promotional code for a new API key

Parameterscoderequired

Version history

1
  1. v1.0.0LatestSep 16, 2026

Score

90% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 daysRequired25 pts
  • Auth declaredRequired10 pts
  • Tools listedRequired15 pts
  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts

    Declare a license in the repository or the registry entry.