mcp-server-madeonsol 1.22.0 → 1.23.0

package/README.md

@@ -13,6 +13,8 @@ MCP server for [MadeOnSol](https://madeonsol.com) Solana KOL intelligence API. U
 
 > Real-time Solana trading intelligence: track 1,069 KOL wallets with <3s latency, score 23,000+ Pump.fun deployers, surface deshred deploy signals **~500ms before on-chain confirmation**, detect multi-KOL coordination, surface bundle-cohort holdings (which same-slot wallets still hold a token's supply), verify any wallet's CURRENT on-chain holdings straight from its token accounts, and stream every DEX trade across 9+ programs. Free tier: 200 requests/day, every endpoint — no signup payment. Get a key at [madeonsol.com/pricing](https://madeonsol.com/pricing).
 
+> **New in 1.23.0 — live holder census: exact holder count, labelled holders, and pools that are named, not just excluded.** The new `madeonsol_token_holders` tool binds `GET /tokens/{mint}/holders` (PRO+): every token account of the mint read from the ledger at `confirmed` and merged per owner, so `concentration.holder_count` is EXACT (distinct non-zero owners minus pools / bonding curves / burns) — never a trade-derived estimate; it is `null` only when the provider refuses the census for a mega-cap, in which case you get the top-20 view and `source.census_fallback_reason` says so. Each disclosed owner carries our labels (`deployer` / `kol` / `early_buyer` / `bundle` / `bot` / `dump_cluster` — empty means unknown to us, not clean), and `excluded[]` NAMES what was taken out of the circulating denominator: `reason` = `pool` (with `dex` + `pool_address`), `bonding_curve` (pump.fun / LaunchLab), `burn`, or `program_account` only when we genuinely cannot attribute the PDA; `pool_pct` / `burned_pct` / `program_pct` split the exclusion. Amounts are raw u64 **strings**. Disclosure: PRO ranks 1–10, ULTRA 1–50, BUSINESS 1–100 — the maths is tier-independent. Big tokens take 5–30 s upstream: you get `503 holder_scan_in_progress` with `retry_after_seconds: 20` while the scan finishes into the cache, and the retry is instant.
+
 > **New in 1.22.0 — two prices on the trade tape, and the right one is now the default.** The trade tape now tells you what a trade actually cost. `price_sol`/`price_usd` on each trade are THIS trade's executed price — `sol_amount / token_amount`, reconciling exactly with the amounts on the same row and with the PnL endpoints. Because `sol_amount` is the wallet's net SOL movement, that is the trader's all-in effective rate: swap fee and any account rent included, not the pool mid. The market-cap tracker's canonical pool price moved to the new **`market_price_sol`/`market_price_usd`** fields — it is sampled once per token per pool update, so every trade in the same slot shares it. Until now `price_sol` carried that canonical value and disagreed with the row's own amounts by a **7.9% median** (p90 ~74%): a stale market price reads low in a pump and high in a dump, so anything you averaged out of the tape inherited the bias instead of cancelling it. Use `price_sol` for cost basis, fills and PnL; `market_price_sol` for a per-token series independent of trade size and direction. Both `madeonsol_token_trades` and `madeonsol_wallet_trades` carry all four fields, and both tool descriptions spell out which price to use so an agent does not average the wrong one.
 
 > **New in 1.21.0 — the Deployer Hunter surface completed.** Seven new operations that existed on the API but had no SDK binding: `madeonsol_deployer_leaderboard`, `madeonsol_deployer_stats`, `madeonsol_deployer_profile`, `madeonsol_deployer_tokens`, `madeonsol_deployer_alert_stats`, `madeonsol_deployer_best_tokens` and `madeonsol_deployer_recent_bonds`. Read `bonding_rate` (lifetime) against `recent_bond_rate` (rolling) — the gap between them is the signal, not either number alone. `runner_rate` only means something once `labeled_tokens >= 3`, and an **untracked wallet returns a profile with zeroed counters, not a 404**, so check `total_deployed` before reading a 0% bond rate as a track record. Dependency ranges are now bounded to the versions actually tested (`@x402/*` `^2.x`, `@solana/kit` `^5.5.1`) instead of open-ended `>=0.0.1`, and the lazily-imported x402 peers are marked optional — a keyed install no longer pulls the whole Solana stack.
@@ -231,6 +233,7 @@ Scored from 1M+ early-buyer records (wallets seen in the first 20 buyers of Pump
 | `madeonsol_token_bundle` | All | Bundle-cohort holdings — which same-slot bundle wallets bought a token and how much of supply they still hold (`held_pct_of_supply` headline, plus `bundle_kind`, `held_ratio`, `fully_exited`). BASIC: bundle block only. PRO: top-10 flags. ULTRA: full cohort + identities |
 | `madeonsol_token_pools` | PRO+ | Per-venue liquidity map — every DEX pool a token trades in (pump.fun/PumpSwap/Raydium/Meteora/Orca) with per-pool `liquidity_usd`, `is_active` (live vs parked), plus a `summary` (pool/DEX counts, `total_liquidity_usd`, `primary_pool`, `top_pool_share_pct` concentration) |
 | `madeonsol_token_depth` | **New 1.20** · PRO+ | Per-pool price impact / slippage — `quotes[]` per SOL buy size (`tokens_out`, `avg_price_sol`, `price_impact_pct`) + `to_move_price` (SOL to move price 1%/5%/10%). `sizes` max 8, default `[0.5, 1, 5, 10]`; unsupported pools (CLMM/DLMM/DBC) flagged with a `reason` |
+| `madeonsol_token_holders` | **New** · PRO+ | Live holder census + concentration — who holds NOW (vs `madeonsol_token_cap_table` = who bought first). `concentration.holder_count` is EXACT (mint-scoped `getProgramAccounts` census merged per owner; `null` only when the provider refuses a mega-cap → top-20 fallback with `source.census_fallback_reason`, never trade-estimated). Each disclosed owner labelled `deployer` / `kol` / `early_buyer` / `bundle` / `bot` / `dump_cluster` (empty = unknown, not clean). Pools / bonding curves / burns EXCLUDED from the circulating denominator and NAMED in `excluded[]` (`reason`: `pool` + `dex` + `pool_address`, `bonding_curve`, `burn`, `program_account`); `amount_raw` is a raw u64 STRING. Disclosure PRO 10 / ULTRA 50 / BUSINESS 100. Big tokens: first call may be HTTP 503 `holder_scan_in_progress` (`retry_after_seconds: 20`) — scan continues + cached, retry is instant |
 | `madeonsol_tokens_batch_risk` | PRO+ | Bulk rug-risk/safety scoring for up to 50 mints — same shape as `madeonsol_token_risk` + `as_of`. Untracked mints return `{ mint, error: "not_tracked" }` without failing the batch; counts as one request |
 | `madeonsol_token_candles` | PRO+ | Historical OHLCV candles (1m–1d). PRO=OHLCV 30d; ULTRA=+net flow, liquidity delta, MEV volume, full history |
 | `madeonsol_token_flow` | PRO+ | Trade-flow aggregate (organic-vs-fake volume) over a 1h/24h `window` — unique wallets/buyers/sellers, buy/sell counts + SOL, `net_sol`, `trades_per_wallet` wash-trading proxy |

package/dist/index.js

@@ -675,6 +675,9 @@ function registerTools(server) {
         server.tool("madeonsol_token_pools", "Per-venue liquidity map for a Solana token — every DEX pool it trades in (pump.fun/PumpSwap/Raydium/Meteora/Orca) with per-pool liquidity, live vs parked (is_active), plus a summary (total liquidity, pool/DEX counts, primary/deepest pool, top_pool_share_pct concentration). Shows WHERE liquidity sits and how fragmented it is, vs the single aggregate number from the token endpoint.", { mint: z.string().describe("Token mint address (base58)") }, { readOnlyHint: true, destructiveHint: false, idempotentHint: true, openWorldHint: true }, async ({ mint }) => ({
             content: [{ type: "text", text: await restQuery("GET", `/tokens/${encodeURIComponent(mint)}/pools`) }],
         }));
+        server.tool("madeonsol_token_holders", "Live holder census + concentration for a Solana token — WHO HOLDS NOW (madeonsol_token_cap_table is who bought first). Read live from the ledger at confirmed: every token account of the mint (mint-scoped getProgramAccounts), merged per owner, ranks 1–100 retained. concentration.holder_count is EXACT — distinct non-zero owners minus the excluded pools/curves/burns — and is null ONLY when the provider refuses the census for a mega-cap (TRUMP/JUP/BONK class): then source.method=getTokenLargestAccounts, source.census_fallback_reason is set and only the top-20 view is served; it is NEVER estimated from trades. Every disclosed owner carries labels[] from MadeOnSol wallet intelligence — deployer / kol / early_buyer / buyer / bundle / bot / dump_cluster (+ kol_name, early_buyer_rank, bot_confidence, historical_win_rate); an EMPTY labels[] means unknown to us, NOT verified clean. Liquidity pools, bonding curves, vaults and burn addresses are EXCLUDED from the circulating denominator and NAMED in excluded[] with reason = pool (dex + pool_address set) | bonding_curve (pump.fun/LaunchLab) | burn | program_account (off-curve owner we could not attribute); concentration splits them into pool_pct / burned_pct / program_pct (over TOTAL supply), while top1/top10/top20/top50/top100_share and deployer/kol/early_buyer/bundle/bot/dump_cluster_pct are over circulating (supply minus excluded). amount_raw / supply_raw / circulating_raw are raw u64 returned as decimal STRINGS — never coerce to a float. Disclosure is tier-gated: PRO ranks 1–10, ULTRA 1–50, BUSINESS 1–100 (the maths is tier-independent). Large established tokens take 5–30 s to enumerate upstream: the first call may return HTTP 503 with error_kind=holder_scan_in_progress and retry_after_seconds=20 — the scan keeps running and is cached, so retry after ~20 s and the answer is instant. 404 not_a_mint = not a mint on-chain; 503 holder_rpc_unavailable (retry 15 s) = we fail closed rather than guess. PRO+ — BASIC receives HTTP 403.", { mint: z.string().describe("Token mint address (base58)") }, { readOnlyHint: true, destructiveHint: false, idempotentHint: true, openWorldHint: true }, async ({ mint }) => ({
+            content: [{ type: "text", text: await restQuery("GET", `/tokens/${encodeURIComponent(mint)}/holders`) }],
+        }));
         server.tool("madeonsol_token_depth", "Per-pool price-impact / slippage for a token — answers 'how much SOL moves this token's price N%' and the impact of each buy size, per pool (NOT router-optimal). Each computable pool returns spot_price_sol, fee_pct, a quotes[] entry per requested SOL size (size_sol, tokens_out, avg_price_sol, price_impact_pct), and to_move_price — the SOL required to move price 1%/5%/10%. Constant-product AMMs are served from stream reserves (source=stream, with reserves_age_ms); pump.fun/bonk bonding curves from a LIVE read of the curve's virtual reserves (source=live_rpc). Pools that can't be priced honestly — concentrated CLMM/Orca/DLMM, Meteora-DBC curves, unclassified models — come back in unsupported_pools[] with a reason (e.g. concentrated_liquidity_depth_not_supported, curve_graduated_use_amm_pool) instead of a wrong number. primary_pool = deepest computable pool; found=false means no pools tracked. PRO/ULTRA only — BASIC receives HTTP 403.", {
             mint: z.string().describe("Token mint address (base58)"),
             sizes: z.array(z.number().gt(0).max(10000)).min(1).max(8).optional()
@@ -1130,6 +1133,7 @@ async function main() {
                         { name: "madeonsol_alpha_wallet", description: "Full alpha profile + bot signals for one wallet. ULTRA only." },
                         { name: "madeonsol_alpha_linked", description: "Behaviorally linked wallets (co-bought 3+ tokens within 2s). ULTRA only." },
                         { name: "madeonsol_token_cap_table", description: "First non-deployer early buyers for a token, enriched. PRO=10, ULTRA=20." },
+                        { name: "madeonsol_token_holders", description: "Live holder census + concentration — who holds NOW. Exact holder_count (null only if the provider refuses a mega-cap), labelled owners, pools/curves/burns excluded and named. PRO=10, ULTRA=50, BUSINESS=100 disclosed; 503 holder_scan_in_progress → retry in 20 s." },
                         { name: "madeonsol_token_buyer_quality", description: "0–100 buyer quality score for a token's first-buyer cohort." },
                         { name: "madeonsol_token_depth", description: "Per-pool price impact / slippage — quotes per SOL buy size + SOL to move price 1%/5%/10%; unsupported pools flagged with a reason. PRO+." },
                         { name: "madeonsol_token_candles", description: "Historical OHLCV price candles (1m–1d). PRO=OHLCV 30d; ULTRA=+net flow, liquidity delta, full history." },

package/dist/version.d.ts

@@ -1 +1 @@
-export declare const VERSION = "1.22.0";
+export declare const VERSION = "1.23.0";

package/dist/version.js

@@ -1,2 +1,2 @@
 // AUTO-GENERATED by ../gen-version.mjs (npm prebuild) from package.json. Do not edit.
-export const VERSION = "1.22.0";
+export const VERSION = "1.23.0";

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "mcp-server-madeonsol",
-  "version": "1.22.0",
+  "version": "1.23.0",
   "mcpName": "io.github.MadeOnSol/madeonsol",
   "description": "MCP server for MadeOnSol Solana KOL intelligence API — use from Claude, Cursor, or any MCP client. Free tier 200 req/day, no payment to start.",
   "type": "module",