KyberSwap Docs

Execute A Classic Swap

Trade Directly Against A Classic Pool

Using the router

We recommend using the router for swapping to and from different assets, which has different swapping functions for ETH, tokens, and tokens that have fee-on-transfer features.
Before executing the swap, it is recommended that an external price source is used to fix the minimum output tokens receivable when selling a fixed amount of tokens, or maximum amount of input tokens to be used when purchasing a fixed amount of tokens.
Your smart contract should also:
  1. 1.
    Have enough ETH and/or tokens when executing the swap
  2. 2.
    Be able to receive ETH, when it is the destination address for swapping to ETH
  3. 3.
    Have granted approval to the router when swapping from tokens

Obtaining pool addresses

It is necessary to specify which pools are to be used for the token swap. Read more about fetching pool addresses before proceeding.
Since each pool represents a token pair, it stands to be the case that the poolsPath specified for token swaps must be 1 size smaller than path, ie. poolsPath.length = path.length - 1. Refer to the examples below.


We will cover 3 scenarios:
  1. 1.
    Swap 1 ETH for DAI
  2. 2.
    Swap USDC to obtain 1 ETH, with WBTC as an intermediary
  3. 3.
    Swap 1 CORE (fee-on-transfer token) for USDT

1 ETH -> DAI


A common error when swapping from ETH is forgetting to actually send ETH when calling the function. The value field should match the amount of ETH sent.
IERC20[] memory path = new address[](2);
path[0] = dmmRouter.weth();
path[1] = dai; // assuming dai is specified as IERC20
// value = 1 ETH, alternatively use msg.value
dmmRouter.swapExactEthForTokens{value: 1e18}(
amountOutMin, // should be obtained via a price oracle, either off or on-chain
poolsPath, // eg. [eth-dai-pool]


Transferring tokens

Before swapping, the contract should be in possession of USDC. The caller can either send the tokens beforehand, or give allowance to the contract to call the transferFrom method. The short code snippet below showcases the latter.
uint256 amountIn = 50 * 10 ** usdc.decimals();
require(usdc.transferFrom(msg.sender, address(this), amountIn), 'transferFrom failed');

Granting Approval

The next step is then to give the router some USDC allowance.
require(usdc.approve(address(dmmRouter), amountIn), 'approve failed');

path and poolsPath

Since the intended token path is usdc -> wbtc -> eth, path = [usdc, wbtc, weth]. We also need to specify the usdc-wbtc and wbtc-eth pools to be used. Deciding which pools to use can be found in this section. Eg. poolsPath=[usdc-wbtc-pool, wbtc-weth-pool], where usdc-wbtc-pool and wbtc-weth-pool are the pool addresses to be used.


Note that if the destination address is a contract, it should have the receive() external payable { ... } or fallback function declaration in order to receive ETH.
IERC20[] memory path = new address[](3);
path[0] = usdc; // assuming usdc is specified as IERC20
path[1] = wbtc; // assuming wbtc is specified as IERC20
path[2] = dmmRouter.weth();
amountOutMin, // should be obtained via a price oracle, either off or on-chain
poolsPath, // eg. [usdc-wbtc-pool, wbtc-weth-pool]
msg.sender, // has to be able to receive ETH


This example is similar to the previous example of USDC -> WBTC -> ETH. The only exception is that CORE is a fee-on-transfer token, and thus requires special handling.


We assume that the previous steps of transferring tokens and token approval to the router has been performed.
IERC20[] memory path = new address[](2);
path[0] = core; // assuming core is specified as IERC20
path[1] = usdt; // assuming usdt is specified as IERC20
amountOutMin, // should be obtained via a price oracle, either off or on-chain
poolsPath, // eg. [core-usdt-pool]