Countersignatory provides on-demand human judgment for AI agents through a paid, auditable marketplace. Request human approval, factual verification, or classification at published prices, pay using USDC on Base or card, and receive a named responder's decision with a reason. Unanswered orders are automatically refunded.
Claude is Anthropic's AI assistant, available as a desktop app and in Cowork. It supports custom connectors, so it can call MCP servers directly from a chat.
This guide connects Countersignatory MCP to Claude Desktop through MewCP, so Claude Desktop can use Countersignatory MCP's 5 tools, like countersignatory_quote, countersignatory_register_interest and countersignatory_spot_index.
Countersignatory MCP needs no credentials, but every call still runs through MewCP: scoped, checked and logged like any other.
More apps
TinyFish MCP for running AI-driven browser automation tasks, deep web research, persistent browser login profiles, remote CDP browser sessions, and agentic wallet payments.
Connect to external APIs, webhooks and internet services through standard HTTP requests. Build integrations with virtually any platform that exposes a web-based interface.
Browse the Apify Store to find pre-built scrapers and automation actors, read their input schemas, run them, and pull the resulting datasets. Includes two ready-made web tools: a search-and-scrape browser that queries Google and returns pages as clean Markdown, and a fetcher that renders JavaScript and gets through bot blocking. Also searches Apify and Crawlee documentation.
AnythingMCP lets AI agents connect to external services through dynamically generated MCP tools. Discover available integrations, configure API connections, and invoke service-specific actions without manually building a separate MCP server for each provider.
Other clients
Countersignatory MCP works the same way in every client below. Pick yours for its own guide.
Free to start.
Scope access to the exact Countersignatory MCP tools and accounts you allow, work and personal side by side. Anything outside that scope is refused at call time, even when the model asks for it.
Each call is logged: what ran, on which account, and whether it worked. Failures come back structured, so the agent knows whether to retry instead of guessing.
Every app sits behind one secure MCP URL, and there is no server for you to deploy or keep running. Add an integration in the dashboard and it is live on the next call, with no config to change and no new credentials to wire up. Tools load only when a task needs them, so your agent stays fast and focused however many apps you connect.
In Claude, open Settings → Connectors, then choose Add custom connector.
Name it "mewcp" and paste the URL.
A browser window opens the first time the client connects. Sign in and approve access. No API key needed.
Start a new chat and ask: "List the tools you can use from MewCP." You should see your connected apps.
Get the current spot price to have a verified human check, agree on, sign off or notarise something. Use when an agent needs human judgment, accountability or a signature, and needs to know the cost and time before committing. Returns an indicative price, a band, what would clear at a lower bid, and the implied rate per human minute.
Record that you would buy. Send the quote_id from a countersignatory_quote response to say the quoted price would clear your use case. For a market on the spot index whose state is opening, which cannot be quoted yet, send its market id instead. This is the demand signal that decides which markets open first.
The public Countersignatory Spot Index: the going rate for verified human judgment by market, with the urgency and off-hours multipliers, the published wage floor, and the derivation of every seeded number. Use it for the reference rate table across all markets; use countersignatory_quote when you have a specific task and need a price for it.
Buy one task from a verified human in a market whose state is live on the spot index: an approval (should this action go ahead), a verification (does this page, document or screenshot show X), or a judgment (which is better, or which label applies). The order is priced at current spot and created unpaid; nothing is charged by this call. The response carries a checkout_url a person can open to pay by card. Once paid it is answered by hand within the SLA, or refunded in full. Read the answer with countersignatory_get_order. Refused at intake: CAPTCHAs and bot checks, logging in or entering credentials, contacting third parties, medical, legal or financial advice, anything sexual or involving minors.
Read one order by the id countersignatory_place_order returned: its status (awaiting_payment, paid, in_progress, answered, refused, refunded or expired), its price, how to pay while it is unpaid, and once answered the answer, the reason, and who gave it and how they are disclosed. A consensus order returns each responder's answer and the majority.
