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

# Introduction

AgentDex is a perpetual-futures DEX built around an MCP server. Bring your own AI agent — Claude, Cursor, Goose, a bespoke framework — point it at one URL with one API key, and trade through the same surface human users do. Per-key trading limits (instruments allowlist, leverage range, order-size range) let you hand a bounded mandate to an agent, so least privilege is the default rather than an afterthought.

### Start here

<table data-view="cards"><thead><tr><th>Title</th><th data-card-target data-type="content-ref">Target</th></tr></thead><tbody><tr><td>New to the venue</td><td><a href="/spaces/aCY4JwuHvGKtRcc70Q10/pages/b5d9ccae65366e3898a6b9a33c18300f13503a04">/spaces/aCY4JwuHvGKtRcc70Q10/pages/b5d9ccae65366e3898a6b9a33c18300f13503a04</a></td></tr><tr><td>Building an integration</td><td><a href="/spaces/aCY4JwuHvGKtRcc70Q10/pages/5452b09fcd912209486fc45d6842f5b20192f4a6">/spaces/aCY4JwuHvGKtRcc70Q10/pages/5452b09fcd912209486fc45d6842f5b20192f4a6</a></td></tr><tr><td>Exploring incentives</td><td><a href="/spaces/aCY4JwuHvGKtRcc70Q10/pages/2106418d32810d9bc14577c78bd88e82877b471c">/spaces/aCY4JwuHvGKtRcc70Q10/pages/2106418d32810d9bc14577c78bd88e82877b471c</a></td></tr><tr><td>Checking trust assumptions</td><td><a href="/spaces/aCY4JwuHvGKtRcc70Q10/pages/79c33e18b015b5909883bd1b79abdae9626d985a">/spaces/aCY4JwuHvGKtRcc70Q10/pages/79c33e18b015b5909883bd1b79abdae9626d985a</a></td></tr></tbody></table>

### Why agent-first

Most exchanges treat the API as a side channel. AgentDex publishes a typed JSON-RPC MCP server alongside REST and WebSocket, with the same authentication, scopes, and limits. The MCP server only exposes the tools a key is allowed to call — a read-only key cannot see `place_order` at all. There is no second permission system, no glue code, no bespoke adapter per LLM client.

The same auth and limits also constrain humans: a delegated team account, a one-off integration, an autonomous bot — all expressed in the same `Permissions` object on a key.

### Underlying design

AgentDex models the venue as a **deterministic state-transition system**: a single pure                  `apply: State × Op → State` function evaluated identically by every honest node. The protocol spec fixes the operation algebra, the state shape, the price reference, and the effect stream observers consume. Two nodes starting from the same committed state and replaying the same ordered log derive the same hash.

This framing has practical consequences:

* The **mark price** has a structural bound on how far any single external observation can move it (η < 1 by construction). Liquidations and margin decisions are robust against single-source manipulation by design — not by ad-hoc safeguards.
* The **canonical log** of operations is what indexers, audit, and market data consume. There is no privileged view.
* On-chain custody lives in **BridgeUpgradeable**, with a force-withdrawal escape hatch (default 7-day timelock) that opens if the off-chain engine ever goes silent.

### Status today, path tomorrow

Honest framing: AgentDex today is a centralised operator running an off-chain matching engine, with on-chain custody and an escape hatch. The Treasury key can authorise bridge balance updates. There are no on-chain fraud proofs yet.

The roadmap moves the venue through four explicit stages of verifiability:

1. **Unverified Data** (today) — structured blocks and transactions written to an internal store.
2. **Validator dry-run** — a clean reference implementation that reproduces engine state from the log; discrepancies logged.
3. **Public log** — blocks published to a blob layer; validator and verification libs open-sourced.
4. **Production core** — the validator implementation becomes the new core; the legacy engine is retired.

We don't pretend any of this is finished. See [Validator and verifiability](/agentdex-docs/under-the-hood/validator-and-verifiability.md) and [Roadmap](/agentdex-docs/under-the-hood/roadmap.md) for what's true today and what's coming, and [Risks](/agentdex-docs/reference/risks.md) for the trust assumptions you are taking on.

### Where to start

<table data-view="cards"><thead><tr><th>Title</th><th data-card-target data-type="content-ref">Target</th></tr></thead><tbody><tr><td>Retail traders — start with Getting started, then Perpetual futures → Fees</td><td><a href="/spaces/aCY4JwuHvGKtRcc70Q10/pages/b5d9ccae65366e3898a6b9a33c18300f13503a04">/spaces/aCY4JwuHvGKtRcc70Q10/pages/b5d9ccae65366e3898a6b9a33c18300f13503a04</a></td></tr><tr><td>Agent / API developers — start with API overview, then MCP server and API keys</td><td><a href="/spaces/aCY4JwuHvGKtRcc70Q10/pages/5452b09fcd912209486fc45d6842f5b20192f4a6">/spaces/aCY4JwuHvGKtRcc70Q10/pages/5452b09fcd912209486fc45d6842f5b20192f4a6</a></td></tr><tr><td>Token hunters — start with Points program and Airdrop scoring</td><td><a href="/spaces/aCY4JwuHvGKtRcc70Q10/pages/2106418d32810d9bc14577c78bd88e82877b471c">/spaces/aCY4JwuHvGKtRcc70Q10/pages/2106418d32810d9bc14577c78bd88e82877b471c</a></td></tr><tr><td>Technical / validators — start with State machine and Validator and verifiability</td><td><a href="/spaces/aCY4JwuHvGKtRcc70Q10/pages/a1e6a0ee9f6d5d23e7262b20698fb098588a68b3">/spaces/aCY4JwuHvGKtRcc70Q10/pages/a1e6a0ee9f6d5d23e7262b20698fb098588a68b3</a></td></tr></tbody></table>

If you only read three pages: this one, [MCP server](/agentdex-docs/developers/mcp-server.md), and [Risks](/agentdex-docs/reference/risks.md).


---

# 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 dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:

```
GET https://agentdex.gitbook.io/agentdex-docs/welcome/introduction.md?ask=<question>&goal=<endgoal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.

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.