> For the complete documentation index, see [llms.txt](https://docs.djzs.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.djzs.ai/readme.md).

# DJZS

**DJZS checks the reasoning behind a trade before the order goes out.**

Send a prediction-market bet or a perp or spot trade in plain words. DJZS returns **PASS**, **WAIT** or **FAIL**, the reasons it fired, the parts it could not settle, and a `verdict_hash` anyone can recompute. It grades whether the idea is stated, falsifiable and bounded. It does not check that your facts are true, and it never says the trade will win.

> **Audit before act.** **Verdict is computed, not improvised.** **Abstain over guess.**

## How a check works

```
your idea, as text
       |
       v
extraction    a model reads the text 3 times and reports facts, never a verdict
       |      (if the 3 readings disagree on a fact, that fact is "unknown")
       v
engine        fixed TypeScript rules: no model, no network, no randomness
       |
       v
PASS / WAIT / FAIL  +  flags  +  unknown fields  +  verdict_hash
```

Same extracted facts, same verdict, same hash, every time (`server/engine-v2/deterministic-engine.ts`). How each part works: [Verdicts](https://docs.djzs.ai/how-a-verdict-is-made/verdicts.md), [Extraction](https://docs.djzs.ai/how-a-verdict-is-made/extraction.md).

## Three ways in

| Who            | Where                                                                                                                                                                 | Cost                                                 |
| -------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------- |
| A person       | Paste the idea at [djzs.ai](https://djzs.ai/)                                                                                                                         | Free, 3 a day per browser                            |
| An agent, free | `POST https://mcp.djzs.ai/try/verdict`: [Free check](https://docs.djzs.ai/for-agents/free-check.md)                                                                   | Free, 3 a day per id                                 |
| An agent, paid | MCP at `https://mcp.djzs.ai/mcp`, or HTTP: [MCP server](https://docs.djzs.ai/for-agents/mcp.md), [Paid check over HTTP](https://docs.djzs.ai/for-agents/paid-http.md) | 2.00 USDC per check on Base, with a permanent record |

The free check and the paid check run the same engine on the same rulesets. The paid check adds the record: an EAS attestation on Base over HTTP, or a Proof-of-Logic certificate on Irys over MCP ([Receipts and certificates](https://docs.djzs.ai/how-a-verdict-is-made/receipts.md)).

## Status, 2026-11-20

Live:

* The free check, in the browser and over HTTP (`djzs-trust-mcp/src/try-route.ts`).
* `verify_pm_trade` and `verify_perp_trade` over MCP (x402) and over HTTP (x402 or MPP), 2.00 USDC per check on Base mainnet (`djzs-trust-mcp/src/index.ts`, `djzs-trust-mcp/src/http-x402-bazaar.v2.ts`).
* Prediction-market ruleset `DJZS-PM-v1.0`: 4 of 4 codes live. Perp and spot ruleset `DJZS-LF-v1.2`: 4 of its 11 codes live (S01, E01, I01, X01). See [Rulesets](https://docs.djzs.ai/how-a-verdict-is-made/rulesets.md).
* Receipts: a paid HTTP check attests on EAS under schema v2. A paid MCP check anchors a certificate on Irys mainnet and, given `agent_address`, writes the agent's on-chain trust score. Both writes fail open: a failed write is reported in the response and never changes the verdict.

Not yet:

* Schema v3, where a resolver contract re-derives the verdict on chain, is built and tested but not deployed. Every DJZS attestation on Base so far uses the v2 schema `0x5ef67e2b8c617635431401d2872b2a6f79eeba54a8e19fd5ecd45aceda2a2030`.
* Whether verdicts predict outcomes is not known. The pre-registered study has 9 of the 100 graded records it needs for a first read. See [What is measured](https://docs.djzs.ai/evidence/measured.md).
* The check reads reasoning only. It does not check that a price is live, that a market is still open, or that a source says what you quote.

***

*Verified 2026-11-20 against commit `1cc11f9` (worker version `c6a886c1`), by probing the live endpoints and by reading attestations on Base. Where these pages and the code disagree, the code wins.*


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://docs.djzs.ai/readme.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.
