MIZAN · Proof of performance
by muaviamohammed · remote · no sign-in to list tools · v1.15.0 · published by GitHub user muaviamohammed
Check before you pay. Prove to get paid. Zero-knowledge proof of what an AI agent earned.
3 tools: 1 do something that can't be undone and 2 change things.
Scanned Oct 21, 2026
- Deletes or cancels0Claude Code: deny
- Can't be undone1Claude Code: ask: see these tools
- Changes things2Claude Code: ask: see these tools
- Only reads0Claude Code: allow
3 checks matched something in the tool definitions. No finding is not a safety guarantee. How the checks work.
claude mcp add --transport http mizan https://studio.mizan.market/mcpFindings
Matches of fixed checks, with the text they matched.
- Infocheck
The description mentions "credential".
What this check looks forMatched text
description: "...our next check returns ALLOW. agent = a credential id, agent id, wallet, ERC-8004 id or ag..."
- Infoprove
The description mentions "credential".
What this check looks forMatched text
description: "...int=true with api_key: a zero-knowledge credential that every agent that checks you will s..."
- Infomore
The description mentions "credential".
What this check looks forMatched text
description: "...se MIZAN does, grouped by job: verify a credential someone shows you, the open standard an..."
Tools
Open a tool for its reasons and full description. Descriptions are the publisher's own words, shown as they are.
3 tools
Can't be undone 1Claude Code: ask
checkTHE ONE CALL BEFORE MONEY MOVES.1 finding
Why it is in this group
- The input carries an amount (input.amount) and a currency, so it moves money.
Findings
- Info
The description mentions "credential".
What this check looks forMatched text
description: "...our next check returns ALLOW. agent = a credential id, agent id, wallet, ERC-8004 id or ag..."
Description, as the server wrote it THE ONE CALL BEFORE MONEY MOVES. Call before you pay, hire, copy, follow or allocate to any agent, bot, wallet or strategy. Returns one signed ALLOW or REFUSE, free, in milliseconds: proceed only on ALLOW. On REFUSE the reply carries a proof request: forward it, and when the counterparty proves itself your next check returns ALLOW. agent = a credential id, agent id, wallet, ERC-8004 id or agent-card URL.
Changes things 2Claude Code: ask
moreEverything else MIZAN does, grouped by job: verify a credential someone shows you, the open standard and its gates, the Bitcoin-anchored log and receipts, your identity and standin…1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
Findings
- Info
The description mentions "credential".
What this check looks forMatched text
description: "...se MIZAN does, grouped by job: verify a credential someone shows you, the open standard an..."
Description, as the server wrote it Everything else MIZAN does, grouped by job: verify a credential someone shows you, the open standard and its gates, the Bitcoin-anchored log and receipts, your identity and standing, checks beyond trading records (jobs, mandates, lending). Every tool listed is callable at this same address; the full list is also at https://studio.mizan.market/mcp/full.
provePROVE WHAT YOUR OWN RECORD REALLY EARNED.1 finding
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
Findings
- Info
The description mentions "credential".
What this check looks forMatched text
description: "...int=true with api_key: a zero-knowledge credential that every agent that checks you will s..."
Description, as the server wrote it PROVE WHAT YOUR OWN RECORD REALLY EARNED. You never reveal your model, code, prompts or strategy: only one position per bar (-1, 0 or +1) on a canonical market leaves your machine. mint=false (default): a free, private verdict with every gate and what would pass. mint=true with api_key: a zero-knowledge credential that every agent that checks you will see. Call with no arguments for the markets and the exact format.
This is not a safety rating, and no finding is not a safety guarantee. The groups and findings come from fixed checks of the tool definitions the server returned on the date shown, not from its code or what it does when called. How the checks work.
Install
Pinned to version 1.15.0. Values in angle brackets are yours to fill in.
claude mcp add --transport http mizan https://studio.mizan.market/mcpRun in a terminal.
Claude Code permissions
Allow the reads, ask before a change, deny the deletes. Paste into .claude/settings.json.
{
"permissions": {
"allow": [],
"ask": [
"mcp__mizan__check",
"mcp__mizan__prove",
"mcp__mizan__more"
],
"deny": []
}
}Tools that only read are allowed, tools that change things ask first, tools that delete or cancel are denied. It assumes the server is installed as mizan. Review it before you use it.
