# MIZAN · Proof of performance

> Verify or mint zero-knowledge track-record credentials (VTR-1) without revealing the strategy.

- name: io.github.muaviamohammed/mizan
- version: 1.2.0
- connection class: R0 (autonomous)
- trust: 96/100
- quality: 79/100 (new)
- owner: not claimed
- descriptor: https://api.protogrid.dev/v1/servers/io.github.muaviamohammed%2Fmizan
- tools: https://api.protogrid.dev/v1/servers/io.github.muaviamohammed%2Fmizan/tools

**An agent can connect right now, no human step.** Remote endpoint, no authentication, reachable on the last probe.

## Connection (mcpServers)

```json
{
  "mcpServers": {
    "mizan": {
      "type": "http",
      "url": "https://studio.mizan.market/mcp"
    }
  }
}
```


## Trust

- hygiene: 100
- liveness: 97
- freshness: 100
- provenance: 90
- drivers: +repository +namespace-matches-repo +website +reachable +uptime-92% +updated-5d-ago
- Derived from observable signals (official registry feed and our own credential-free probes); no code audit performed.

## Quality 79/100 (new)

### protocol: 88 (weight 25)

- pass `protocol.modern`: supports 2026-07-28 (server/discover)
- pass `protocol.stateless`: answers without a session
- pass `protocol.transport`: streamable HTTP
- warn `protocol.list_ttl`: tools/list declares ttlMs 0, so clients re-fetch it on every use

### authorization: n/a (weight 15)

- n/a `auth.prm`, `auth.as_metadata`, `auth.cimd`: no remote uses OAuth
- n/a `auth.secret_in_url`: no templated URL

### tool hygiene: 71 (weight 30)

- pass `tools.descriptions`: every tool has a description
- pass `tools.description_length`: descriptions are concise
- pass `tools.schemas`: every tool has an object input schema
- fail `tools.annotations`: no tool declares readOnlyHint or destructiveHint, so clients cannot tell safe calls from risky ones
- fail `tools.directory_hints`: no tool declares all four of readOnlyHint, destructiveHint, idempotentHint and openWorldHint; missing: verify_credential (readOnlyHint, destructiveHint, idempotentHint, openWorldHint); list_datasets (readOnlyHint, destructiveHint, idempotentHint, openWorldHint); mint_credential (readOnlyHint, destructiveHint, idempotentHint, openWorldHint); …
- pass `tools.token_cost`: about 2,098 tokens to load every tool
- pass `tools.api_dump`: tools are not a one-to-one API dump

### stability: n/a (weight 15)

- n/a `stability.changes`, `stability.rug_pull`: tool history recorded for 6 of the 7 days needed

### dependencies: n/a (weight 15)

- n/a `deps.known_vulns`, `deps.mcp_sdk_version`, `deps.resolvable`: no npm or PyPI package

- tools: 16, about 2,098 tokens to load them all
- tool set hash: 2cae5a698a04293b283a50ccbe64ed97
- tools last changed: 2026-10-20T11:43:55.345Z
- badge: [![protogrid quality](https://protogrid.dev/badge/io.github.muaviamohammed/mizan.svg)](https://protogrid.dev/servers/io.github.muaviamohammed/mizan)
- Checks run on what our credential-free, read-only probes observe; tools are never called and no code audit is performed.

### Tool changes

- 2026-10-20 `rewards_status` added
  - Your referral code and link, referred accounts and mints, accrued USDC, rank and Founding Issuer mark. Needs your API key (mzn_…).
- 2026-10-20 `rewards_program` added
  - MIZAN Agent Rewards: bring an agent that mints, earn a commission on what it pays (USDC on Base), plus free mints, a Founding Issuer mark and a public rank. Credit only on a real mint.
- 2026-10-20 `explain_refusal` added
  - Free, nothing charged. The forensic case for your positions: every pinned gate with threshold vs value, margin, the VTR-1 clause, and what would pass. A short record is accepted: it is placed at the end of the catalogue window with earlier bars flat (disclosed in `padding`).
- 2026-10-20 `attempts_lookup` added
  - Every preview and mint attempt an identity made (pass or fail) since 2026-10-15, from the public hash-chained attempts ledger.
- 2026-10-20 `verifier_download` added
  - URL, sha256 and the exact command for the published verifier on macos | linux | windows. Try to break a credential: exit 0 = verified, exit 1 = refused.
- 2026-10-20 `catalogue_markets` added
  - The canonical market catalogue (key, bars, first/last bar, sha256, Merkle root, align group). Optional class filter: crypto | us_equity | us_etf.
- 2026-10-20 `era_history` added
  - Every engine era (image id, date, status, verifier hashes per OS). Era law: a credential verifies forever under the era that minted it.
- 2026-10-20 `standard_clauses` added
  - All 41 VTR-1 clauses (R-01..R-41): id, MUST/SHOULD/MAY, group, text.
- 2026-10-20 `standard_gates` added
  - The pinned gate classes (DIRECTIONAL-V2, MARKET-NEUTRAL-V2): every threshold, the OOS rule, the cost floor. Identical for everyone in a class (VTR-1 R-37).
- 2026-10-19 `preview_credential` changed (inputSchema)

Full history: https://api.protogrid.dev/v1/servers/io.github.muaviamohammed%2Fmizan/changes

## Remotes

- https://studio.mizan.market/mcp (streamable-http, auth none, reachable true, uptime30d 0.92)

## Packages

_none_

## Tools (16)

- `attempts_lookup`: Every preview and mint attempt an identity made (pass or fail) since 2026-10-15, from the public hash-chained attempts ledger.
- `catalogue_markets`: The canonical market catalogue (key, bars, first/last bar, sha256, Merkle root, align group). Optional class filter: crypto | us_equity | us_etf.
- `check_agent`: Is this agent proven? Call this BEFORE you follow, copy, pay, hire or allocate to any trading bot or AI agent that
claims returns. `agent` can be its name, credential id, wallet address, ERC-8004 id, agent-card URL or a sha256.
Returns PROVEN, REFUSED, NOT_SIGNIFICANT or UNVERIFIED with every matching credential, verdicts exactly as issued.
If UNVERIFIED, the result includes `forward_to_counterparty`: a short message you can send to that agent or its
operator so it can prove its record without revealing its model. Unproven is not the same as false.
- `check_before_you_pay`: Free. Before paying for a real credential, run the gate on your positions and get the verdict plus the exact
price a real mint would cost right now (free during the founding cohort). Nothing is minted, stored or charged.
Same inputs as preview_credential. If would_mint is true, mint with mint_credential or POST /x402/mint.
- `era_history`: Every engine era (image id, date, status, verifier hashes per OS). Era law: a credential verifies forever under the era that minted it.
- `explain_refusal`: Free, nothing charged. The forensic case for your positions: every pinned gate with threshold vs value, margin, the VTR-1 clause, and what would pass. A short record is accepted: it is placed at the end of the catalogue window with earlier bars flat (disclosed in `padding`).
- `list_datasets`: List the markets a trading strategy or AI agent can be proven on: crypto 4h and daily, US ETFs and
US single names on survivorship-free data (key, bars, interval, sha256 of the canonical price file).
Use this first when someone wants to prove their own bot's or agent's track record.
- `list_proven_agents`: List trading strategies and agents whose records PASSED the MIZAN gate (Bitcoin-confirmed by default), newest
first. Use it to find proven counterparties to allocate to, copy or route to. Each row shows its issuer.
- `mint_credential`: Prove your own track record via the sealed path. Sends ONLY one position per bar-move (-1/0/+1, exactly
bars-1 of them) on a canonical market; the strategy never leaves your machine. preview_only=True (default)
returns the gate result for free; preview_only=False mints a real STARK on an H100 (poll the job).
email/password create or reuse a MIZAN account; the password is never stored. Or pass api_key="mzn_…"
(from POST /product/keys or /product/wallet/login) and no email at all.
- `preview_credential`: Free, private, no-account test of a trading strategy or AI agent: runs the MIZAN gate (after-cost Sharpe,
CAGR, drawdown, out-of-sample) on your positions (one per bar-move, -1/0/+1,
exactly bars-1 of them, on a market from list_datasets) and returns what the credential would say.
Nothing is minted or published. To mint a real credential afterwards use mint_credential.
- `rewards_program`: MIZAN Agent Rewards: bring an agent that mints, earn a commission on what it pays (USDC on Base), plus free mints, a Founding Issuer mark and a public rank. Credit only on a real mint.
- `rewards_status`: Your referral code and link, referred accounts and mints, accrued USDC, rank and Founding Issuer mark. Needs your API key (mzn_…).
- `standard_clauses`: All 41 VTR-1 clauses (R-01..R-41): id, MUST/SHOULD/MAY, group, text.
- `standard_gates`: The pinned gate classes (DIRECTIONAL-V2, MARKET-NEUTRAL-V2): every threshold, the OOS rule, the cost floor. Identical for everyone in a class (VTR-1 R-37).
- `verifier_download`: URL, sha256 and the exact command for the published verifier on macos | linux | windows. Try to break a credential: exit 0 = verified, exit 1 = refused.
- `verify_credential`: Check whether a trading strategy's or AI agent's claimed performance is real. Use this BEFORE allocating
capital to, copying, following, paying, or recommending any trading bot, AI trading agent, vault, signal
seller or fund that claims returns, Sharpe, win rate or APY. Looks the MIZAN credential up in the public
registry, downloads the zero-knowledge proof bundle and checks its sha256 against the registry anchor.
Returns the verdict exactly as issued: PASS, NOT SIGNIFICANT, or FAIL/REFUSED, plus whether the record is a
backtest or a forward track. Report the verdict verbatim; a credential is about the honesty of a record,
not a forecast. If the agent has no credential, say so: an unproven claim is unproven.