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 works best when tools are presented in clear domains rather than as one giant bucket of magical functions.

​
The five categories

​
1. Asset and market tools

Identify assets and fetch market context. Typical examples: token lookup, token metadata, token price, amount conversion, chain or network context, trending tokens, news and sentiment. → See Market Data for the concrete tools we ship.

​
2. Wallet and portfolio tools

Inspect what a wallet currently holds or is doing. Typical examples: wallet balances, position lookup, portfolio summary, transaction history, health and risk indicators, watchlists and alerts. → See Wallet & DeFi.

​
3. Routing and execution tools

Move from idea to executable plan. Typical examples: swap quote, bridge quote, pool or vault details, build transaction, simulate transaction, perform transaction, gas estimate. → See Wallet & DeFi (ZapCallDataGeneratorTool, PerformZapTool).

​
4. Safety and policy tools

Validate actions before anything touches real execution. Typical examples: token verification, approval checks, allowance checks, route sanity validation, slippage policy checks, execution risk evaluation. → Surfaces today through SDK approval modes; richer policy tools land with the runtime layer (roadmap).

​
5. Knowledge and support tools

Help assistants answer integration or product questions. Typical examples: docs retrieval, RAG over documentation, integration guidance, troubleshooting support. → See Data Sources (RagOverDocsTool, WebSearchTool).

​
How tools work

Each tool has:
  • Name — identifier the agent uses to call it.
  • Description — natural-language hint for the model.
  • Parameters — typed schema.
  • execute — async function that does the work.
When ZapBot decides to call a tool, the model emits structured arguments matching the schema; the runtime validates, calls execute, and feeds results back.

​
Read vs execution-sensitive

ReadExecution-sensitive
PriceTool, BalanceTool, PoolTool, WebSearchTool, …PerformZapTool, ZapCallDataGeneratorTool, ChangeChainTool
Always safeConfirmation required
Cached where possibleIdempotent on retry; no double-execution
For the safety model, see Safety & Execution.

​
Suggested maturity model

CategoryBest first stepLater expansion
Asset & marketprice, token lookupnews, sentiment, analytics
Wallet & portfoliobalances, positionsalerts, history, health monitoring
Routing & executionquotes, build, simulatepolicy-aware execute
Safety & policyslippage, validationapproval and risk engines
Knowledgedocs searchproduct copilots and support agents

​
Calling tools directly

You don’t have to go through the agent. From SDK-AI: 
const priceTool = sdk.getTool('PriceTool');
const result = await priceTool.execute({
  tokenAddresses: '0xA0b86991c6218b36c1d19d4a2e9eb0ce3606eB48',
  chainId: 1,
});
Direct calls skip the LLM — faster, cheaper, deterministic.

​
Concrete tool inventory

Tools shipped today (from ToolNames.ts):
  • PriceTool, BalanceTool, TokenAddressTool, DollarToTokenAmountTool
  • TrendingNewsTool, TrendingTokenTool, SentimentTool, PricePredictionTool
  • ZapCallDataGeneratorTool, BridgeLinkGeneratorTool, SwapLinkGeneratorTool
  • WebSearchTool, DefiPositionsTool, RagOverDocsTool, PoolTool
  • RequestChainChangeTool, ChangeChainTool, ShowBalanceUITool
  • SheluderTool, GetAllSchedulesTool, PerformZapTool
For per-tool parameters and examples, see the four detail pages:

Wallet & DeFi

Balances, pools, positions, zap execution.

Market Data

Prices, predictions, sentiment, trending.

Data Sources

Web search, RAG, external APIs.

Utilities

Link generators, scheduler, chain change.