Pro and above · Compare plans and subscribe
Build a live Solana bundle monitor
Track coordinated Solana buys and exits over WebSocket, deduplicate wallet exposure and distinguish proven bundle holdings from same-slot estimates.
By MadeOnSol · Updated
Integration choice
Treat a bundle event as a trigger to refresh the affected mint. Use the bundle-share response for current holdings and keep its observation time, evidence basis and scope beside the percentage.
Before you start
- Use one mint, a fixed observation window and server-side credentials.
- Log event identity and the measurement’s as_of separately so replay is testable.
Implementation steps
- Set SOLANA_MINT to a traded mint. Load bundle-share with an explicit comparison window such as 24h. Keep the same window and min_wallets across requests.
- Watch token:bundles for the mints in view. On a new cluster or exit, mark its cached reading dirty. Coalesce refreshes by mint, window and threshold across viewers.
- Keep the strict lower bound and the broader same-slot figure in separate fields. Replace a successful reading atomically. Preserve null and the returned scope when balance evidence is incomplete.
- Deduplicate repeated event deliveries before issuing notifications. After recovery, refresh the current measurement once and keep historical exit events in their own timeline.
What does proven bundle ownership measure?
The strict percentage measures current supply held by wallets linked through the supported launch-bundle or same-transaction evidence. It is a lower bound. The broad percentage also considers same-slot buying. Independent buyers can share a busy slot, so a scanner should retain that distinction in its labels.
Map strict_bundle_pct_of_supply_now to the proven lower bound and bundled_pct_of_supply_now to the broader reading. Store bundled_pct_basis, scope and as_of with the numbers. A later response can have different evidence even when the displayed percentage stays the same.
Work through an overlapping-wallet example
Suppose the first cluster contains A and B, and a later one contains B and C. At one observation their balances are 30, 20 and 10 out of 1,000 units. For that illustrative cohort the distinct-wallet share is 6%. Adding the cluster totals produces 8% by counting B twice. Historical buy amounts answer a different question from current balances.
Handle a sell alert inside the cache interval
An exit milestone describes sales by tracked wallets. A balance may also leave through a transfer. Keep the latest measured share and its as_of on screen while refreshing. The standard share read is cached for 30 seconds, so repeated reads in that interval can legitimately return the same observation.
Code examples
Subscribe on an authenticated Node.js socket
const token = process.env.SOLANA_MINT;
if (!token) throw new Error("Set SOLANA_MINT first");
socket.send(JSON.stringify({
type: "subscribe",
sub_id: "solana-live-bundle-monitor",
channels: ["token:bundles"],
filters: { mints: [token] }
}));Read current holdings for the 24-hour window
curl -H "Authorization: Bearer $MADEONSOL_API_KEY" "https://madeonsol.com/api/v1/tokens/$SOLANA_MINT/bundle-share?window=24h&min_wallets=3"Common mistakes and fixes
| Mistake | Recommended fix |
|---|---|
| A transfer is presented as a sale. | Require an exit event for a sale milestone. Use share measurements for current exposure. |
| A restricted reading looks like full coverage. | Carry scope, evidence basis and missing values through storage and rendering. |
Verify your integration
Feed the same event twice, then a second cluster containing a previously seen wallet. Expect one alert for the duplicate and one shared refresh per cache interval. Simulate a null share and a launch-cohort-only response and check their labels.
Production checklist
- Duplicate events produce one alert.
- The query window is displayed.
- Strict and broad percentages have separate labels.
Frequently asked questions
Can I add launch and later cluster percentages?
Wallets can belong to both. Use the deduplicated share response for the selected population.
Does a null percentage mean every wallet exited?
No. A null measurement can reflect missing balance evidence. A measured zero and an unavailable value require different handling.
