Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.dzap.io/llms.txt

Use this file to discover all available pages before exploring further.

DZap AI is the developer-facing layer for connecting LLMs, agent frameworks, and client applications to DZap capabilities. Build natural-language DeFi assistants, portfolio agents, internal copilots, and execution-aware AI products on top of DZap infrastructure — through a clean tool surface.

​
What this is for

Teams building:
  • Natural-language DeFi assistants
  • Portfolio and monitoring agents
  • Internal copilots for strategy research
  • Developer tools and MCP-enabled workflows
  • Execution-aware AI products on top of DZap

​
Why DZap AI

Unified DeFi access

Work against a single tool surface instead of stitching together quotes, balances, routes, and execution helpers by hand.

MCP-ready integration

Connect from MCP-compatible clients and agent frameworks with a familiar tool-based interface.

Separation of concerns

Keep deterministic protocol actions separate from reasoning — the layer where mistakes get expensive.

Safety-aware design

Treat execution, permissions, simulation, and validation as product features, not afterthoughts.

​
Platform model

┌──────────────────────────────────────────────┐
│   Interfaces — MCP, CLI, Apps, Plugins       │
└─────────────────────┬────────────────────────┘
                      │
┌─────────────────────▼────────────────────────┐
│   Runtime & Permissions                      │
└─────────────────────┬────────────────────────┘
                      │
┌─────────────────────▼────────────────────────┐
│   Reasoning / Agent Layer                    │
└─────────────────────┬────────────────────────┘
                      │
┌─────────────────────▼────────────────────────┐
│   Tool Layer                                 │
└─────────────────────┬────────────────────────┘
                      │
┌─────────────────────▼────────────────────────┐
│   Protocol Capabilities                      │
└──────────────────────────────────────────────┘
  • Protocol capabilities — quotes, positions, transaction building, simulation, execution primitives.
  • Tooling — wraps capabilities with descriptions, schemas, and safer contracts for AI use.
  • Reasoning — intent, multi-step planning, tool orchestration.
  • Runtime — sessions, approvals, execution boundaries.
  • Interfaces — MCP, chat surfaces, apps, developer workflows.
See Architecture for the full breakdown.

​
Common use cases

AI portfolio assistant

Read balances, surface positions, monitor changes, route follow-up actions into DZap tooling.

Strategy copilot

Combine market data, protocol context, and route generation to help users move from intent to execution.

Internal ops + support

Give teams a structured way to query docs, inspect routes, look up tokens, and assist with debugging or onboarding.

MCP integrations

Use DZap AI from Claude Desktop or custom internal clients via the standard MCP flow.

​
Documentation map

Quickstart

First integration: read-only or execution-aware.

Architecture

The five-layer system in detail.

MCP Server

Connect via https://ai.dzap.io/sse.

Tool Categories

Asset · Wallet · Routing · Safety · Knowledge.

Safety & Execution

What’s read-only vs gated.

Agent Integration

llms.txt, OpenAPI, plugin manifest.

​
Current focus

The current public developer story is strongest around AI-accessible DeFi tools, MCP integration, and structured execution workflows. Broader runtime, policy, and automation surfaces grow on that foundation without breaking the mental model.