> ## Documentation Index
> Fetch the complete documentation index at: https://docs.nex-t1.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Tools

> Built‑in and MCP adapter tools, and how to add your own.

# Tools

## Built‑in

* DuckDuckGo search tool included and bound to the LangGraph agent by default.
  * Location: `app/core/langgraph/tools/duckduckgo_search.py`
  * Registration: `app/core/langgraph/tools/__init__.py` (edit `tools` list)

## MCP Adapters (selected)

* Research: The Graph (`research_thegraph.py`), Dune (`research_dune.py`), CoinGecko (`research_coingecko.py`), DeFiLlama helpers via integrations.
* EVM: Chainlink (`evm_chainlink.py`), Uniswap/0x (`evm_uniswap.py`), Thirdweb Wallet (`evm_thirdweb.py`).
* Exchanges: Binance typed helpers (`integrations/mcp/binance_helpers.py`).
* Solana: Jupiter quote/execute (`sol_jupiter.py`).
* Bitcoin: Bitcoin MCP adapter (`integrations/mcp/bitcoin.py`).

These adapters support MCP servers via env vars and often include direct HTTP fallbacks for development.

## Add a New Tool

1. Create a `BaseTool` subclass in `app/core/langgraph/tools/`.
2. Export it from `app/core/langgraph/tools/__init__.py` and add to the `tools` list.
3. Optionally expose a dedicated endpoint that uses the tool for easier user access.

<Info>
  To bind MCP adapters as agent tools, import and add them to the `tools` list. Ensure inputs/outputs are compact strings or structured content that fits the chat flow.
</Info>