For the complete documentation index, see llms.txt. This page is also available as Markdown.

Integration recipes

Short, copy-pasteable examples for the most common things integrators and agents do on Kumbaya. All of these are grounded in the actual API and contract surface - no pseudocode.

Building an LLM agent rather than a script? Most of these flows already exist as one-call tools and skills in the Kumbaya Agent Kit - e.g. buying a token (recipe 3) is the swap tool, and sweeping creator earnings (recipe 4) is claim_fees + withdraw_tips. Use these recipes when you want the raw calls; use the kit when you want the model to drive.

1. Watch for new launches (live)

Subscribe to the indexer and react as new tokens are deployed.

import { createClient } from 'graphql-ws'
import { WebSocket } from 'ws'

const client = createClient({
  url: 'wss://ql.kumbaya.xyz/v1/graphql',
  webSocketImpl: WebSocket,
})

const unsubscribe = client.subscribe(
  {
    query: `
      subscription NewLaunches($chainId: Int!) {
        FireToken(
          where: { chainId: { _eq: $chainId } }
          order_by: { createdAt: desc }
          limit: 10
        ) {
          address creator createdAt
          pool { id }
          marketCapUSD graduationProgress
        }
      }
    `,
    variables: { chainId: 4326 },
  },
  {
    next: ({ data }) => {
      const latest = data?.FireToken?.[0]
      if (latest && Date.now() / 1000 - Number(latest.createdAt) < 60) {
        console.log('New launch:', latest.address)
        // your hook: e.g. evaluate, snipe, post a tweet
      }
    },
    error: console.error,
    complete: () => console.log('subscription ended'),
  },
)

2. "Heating up" - find tokens with rising volume

The indexer tracks 30m/1h/2h-prior buy and sell counts plus velocity ratios. This is exactly how Kumbaya's "Heating up" feed is computed.

3. Buy a launchpad token - quote then swap

Use the Exchange API for routing; submit calldata yourself.

For native ETH ↔ token swaps the API handles wrapping for you. For pre-approved ERC-20 → ERC-20, do the approve once, then run the swap.

4. Sweep your creator earnings

For a graduated launchpad token where you (or an address you control) is the creator:

Run on a schedule (e.g. daily) for any token you've launched.

5. Find every token I created (and what they've earned)

6. Trigger graduation on tokens that are ready

After gracePeriodDuration elapses (read live from FireRegistry.gracePeriodDuration() - see Live registry config), call FireGraduator.graduate(token) to actually finalize.

7. Search-as-you-type token picker

Build a token-picker UI without running your own index:

For full search results (tokens + pools, multiple matches):

Public, no auth, rate-limited per IP.

8. Newest swaps for a pool (REST, no GraphQL)

Where poolId is the address-chainId composite (e.g. 0xabc...-4326).

9. Read protocol-level config live

Don't hardcode tunables. Read on demand:

See Live registry config for the full field list.

10. Build a creator-fee dashboard

Combine indexer + on-chain reads:

FireStream doesn't expose a "preview pending fees" view - to know what the next claimFees would pay out, you'd simulate it (e.g. with eth_call) or just call it and parse the BeneficiaryPaid event from the receipt.

More

Last updated