Build a discovery board

A discovery board follows launchpad tokens through their life: just launched, close to graduating, and graduated. POST /v1/discovery returns all three columns in one request.

1. Request the columns

curl "https://{{API_HOST}}/v1/discovery" \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "chains": ["solana"],
    "newPairs": {
      "limit": 30,
      "filters": { "withAtLeastOneSocial": true, "devHoldingPercent": { "max": 10 } }
    },
    "graduating": {
      "limit": 30,
      "filters": { "holderCount": { "min": 100 }, "maxGraduationStaleMinutes": 30 }
    },
    "graduated": {
      "sortBy": "volume24hUsd",
      "limit": 30,
      "filters": { "liquidityUsd": { "min": "10000" } }
    }
  }'

Each column takes its own sortBy, order, limit and filters. Every filter and sort field is documented in Filtering and sorting. The graduating and graduated columns also apply minimums of their own, such as 10 holders. Your filters can raise these minimums but not lower them.

2. Render the response

{
  "columns": [
    { "column": "newPairs", "deltas": [ { "chain": "solana", "address": "…", "rank": 1, "data": { } } ] },
    { "column": "graduating", "deltas": [ ] },
    { "column": "graduated", "deltas": [ ] }
  ]
}

Each column’s deltas is the whole ranked list. Draw the items in rank order. data is the full token object, so each card can show:

Card element	Field
Name and symbol	`data.name`, `data.symbol`
Age	now minus `data.createdAt`
Market cap	`data.marketCap.usd`
Curve progress	`data.bondingCurvePct`
5-minute volume and trades	`data.stats["5m"].volume.usd`, `data.stats["5m"].buyCount`, `data.stats["5m"].sellCount`
Holders	`data.holders.count`
Risk badges	`data.holders.devPct`, `data.holders.sniperPct`, `data.holders.bundlerPct`, `data.security.mintDisabled`
Links	`data.metadata.twitter`, `data.metadata.website`, `data.metadata.telegram`

3. Keep it fresh

Poll the endpoint on an interval. Every 2 to 5 seconds suits a fast-moving newPairs column, and your request budget leaves plenty of room. When you redraw, match cards by chain + address, so a token that moves between ranks animates instead of flickering. A token that has left a column’s ranked list is gone from that column; a token that graduated shows up in graduated. Now and then a card’s windowed stats read zero for a response while the token’s full record catches up. Keep the previous values for a card when that happens, and the next refresh corrects it.

Filter presets

Preset	Filters
Hide likely rugs	`devHoldingPercent: { max: 10 }`, `snipersSupplyPercent: { max: 20 }`, `top10HoldersPercent: { max: 40 }`
Socials required	`withAtLeastOneSocial: true`
Only active	`lastActivityMinutes: 5`
Real liquidity	`liquidityUsd: { min: "5000" }`
One launchpad	`protocols: ["pumpfun"]`
No copycats	`excludeSymbolOrName: ["test", "scam"]`

Tune the numbers to your users; these are starting points, not recommendations.

Several chains

Send more than one slug in chains, for example ["solana", "base", "bsc"]. Each column then ranks tokens from every listed chain as one set. Use data.chain to badge each card.

Get started

Core concepts

Streams

Recipes

Resources

Build a discovery board

1. Request the columns

2. Render the response

3. Keep it fresh

Filter presets

Several chains

​1. Request the columns

​2. Render the response

​3. Keep it fresh

​Filter presets

​Several chains

1. Request the columns

2. Render the response

3. Keep it fresh

Filter presets

Several chains