---
name: rh-launchpad-sector
description: The Robinhood Chain launchpad sector mapped by protocol revenue, not TVL — every pad after Pons, which have tokens, which capture fees, and the verified burn numbers. Read before answering any "what's the next Pons" or "best launchpad on RH" question.
sources: [cowork]
aliases: [robinhood-launchpads, next-pons, rh-pads]
---

# Robinhood Chain launchpad sector — worked 2 Sep 2026

Companion to [[rh-chain-thesis]], [[rh-chain-infrastructure]]. Ben asked: best launchpads after Pons, and which are the best investments.

## THE SHAPE: it is a monopoly, not a sector
48 protocols carry the Launchpad tag on this chain. Revenue 24h (2 Sep, api.llama.fi/overview/fees/robinhood?dataType=dailyRevenue):
Pons V2 $1,035,295 · o1 Launchpad $268,286 · Pons V1 $71,142 · StonkBrokers $22,374 · Bags $11,077 · Flap $9,177 · LetsCash $6,721 · PAIR $6,341 · Coinbarrel $2,315 · Sentry $1,700 · token.select $746 · Sushi Launchpad $676 · Hookers $696 · bow.fun $162. Everything else is zero.
**Pons V1+V2 = $1,106,437/24h = 76% of all launchpad protocol revenue on the chain. The #3 with a token does 2%.**
Gross fees 24h tell the same story: Pons V2 $5,638,023 · o1 $499,835 · Pons V1 $335,884 · NOXA $153,498 · LetsCash $57,569 · Pools $38,904 · StonkBrokers $32,087 · Bags $22,155 · PAIR $21,008.

## PONS IS CHEAPER THAN EVERYTHING BEHIND IT — the finding that decides the question
- PONS 0x39dbed3a2bd333467115de45665cc57f813c4571, mcap $275.6M at $0.3898.
- **BURN VERIFIED ON CHAIN: 293,199,140 of 1,000,000,000 sits at 0x...dead = 29.32% of supply, in about two months.** DexScreener already nets this out, so $275.6M is the burn-adjusted cap; no correction needed.
- Lifetime protocol revenue V1 $6,164,874 + V2 $6,007,424 = $12,172,298. 80% burned = ~$9.7M spent to retire 293.2M tokens, so **average burn price ~$0.033, i.e. 11.7x below spot.** The burn is real, it was cheap, and it worked. This is the opposite of the HOOKR burner ($2,348 lifetime).
- Run-rate: $1.106M/day = $404M/yr, cap/rev **0.68x**, burn 117% of cap per year. Trailing 30d $6.90M = $83.9M/yr, cap/rev **3.28x**.
- pump.fun comp 2 Sep: PUMP mcap $4.24B, revenue 24h $1,326,114, 30d $36,462,123 = ann $444M, **cap/rev 9.5x**. Pons does 83% of pump.fun's daily revenue at 6.5% of its cap.
- Only question remains step-change vs spike. Daily rev V2: 20 Aug $31.6k -> 25 Aug $141k -> 29 Aug $550k -> 31 Aug $865k -> 2 Sep $1,035k. Thirteen days, 33x, no down day since 28 Aug.

## THE CANDIDATES AFTER PONS, ranked
1. **PAIR / pair.fund 0x6b1d42927B1a84eC28Fa88d4fC6FA7AF404966be** — mcap $4.9M, liq $263,077, v24 $1.55M. THE ONLY POST-PONS PAD WITH A TOKEN THAT VERIFIABLY BURNS. Multi-pool launches paired against 1-5 Robinhood Stock Tokens at once on Uniswap v4; ABI confirms it (`launchTokenMulti`, `allocations[] {quoteToken, weightBps}`, `quoteStockToken`, `claimCreatorFees`, `migrate`). **Burn CONFIRMED on chain: 83,129,220 = 8.31% of supply at 0x...dead.** Token bytecode has NO mint, pause, owner, blacklist, setFee or setMaxWallet selectors: clean and immutable. Own Dune-backed API https://pair.fund/api/stats/protocol (methodology published, query 8520127): rolling 24h volume $7,872,323, 94 launches, 32,070 trades, avg trade $257, **net flow MINUS $101,980**, top-3 market concentration 88.7% (AMC $3.84M + SPY $3.00M of $7.87M). PAIR v2 teased 2 Sep.
2. **o1 Launchpad** — the only pad whose RH revenue is genuinely accelerating: RH-chain revenue 29 Aug $439 -> 30 Aug $6,736 -> 31 Aug $74,333 -> 1 Sep $246,935 -> 2 Sep $268,286, while its Base revenue sat flat at ~$21k. Take rate 53.7% of fees vs Pons's 18.4%. **BUT THE TOKEN DOES NOT CAPTURE IT.** $O 0x182FA643E5f29d5EcA75e7b9CF9336A3fe4620b2 is on BASE, not RH. Whitepaper: utility is fee discounts and early feature access; the only revenue language is that stakers "may become eligible to claim limited ecosystem badges" which "qualify to APPLY for a portion of platform revenue sharing" and "do not confer any enforceable rights". No buyback, no burn. mcap $87.2M / FDV $545.1M = 16% float, unchanged since TGE 17 Jun 2026; investors 18% + team 10% on 1-year cliff, so the first cliff is June 2027. Backed by Coinbase Ventures, a16z, AllianceDAO, Amber, $4.8M seed; on the Coinbase listing roadmap.
3. **StonkBrokers / STONKBROKER 0xe934e36a439c94017b64a3fece66af12099abf50** — mcap $34.2M (was $70.9M on 30 Aug, -52%), rev $22,374/24h = $8.2M/yr, cap/rev 4.2x. 30d change -51.9%. Burn at dead: 1,188,272 of 2,392,625,319 = 0.05%. Revenue peaked 20 Aug at $250,720/day and has run $5k-26k since. Ben holds $12,900.
4. **LetsCash / CASHCAT 0x020bfc650a365f8bb26819deaabf3e21291018b4** — mcap $272.4M against $6,721/day of launchpad revenue = $2.45M/yr, **cap/rev 111x**. Burn 11,827,026 = 1.18%. Priced as a meme, not as the pad it fronts. Worst risk-reward in the sector.
5. **Pools / pools.trade** — $1,514,075 of 30d gross fees but **$0 protocol revenue**: no take rate, so nothing can accrue. Two tokens off the same site, FRONG 0x6245e67affA44a23077f0Ea7f981a8DC743a0c47 mcap $10.7M (burn 0.60%) and POOLS 0x385b36Ff682Ab4C76E7c37A66b96aABC466471d5 mcap $2.36M (burn 138 tokens, zero). Trading vehicles, not revenue claims.
6. Dead or not investable: **NOXA** $4.22M of 30d fees but $0 revenue and no token, and it stopped accepting new launches mid-July 2026 — those fees are residual trading on legacy tokens like CASHCAT, it is the corpse Pons ate, not a candidate. **Flap** is BSC-native ($340k/day BSC vs $9k RH). **Bags** is Solana-native, no token, its RH "BAGS" tickers are $22k impersonators. **Sentry** 0x1EcA20cfa4AF2e2fA2F4CE2bF8d97bFa184FD4D7 mcap $4.08M on $1,700/day. **token.select / SELECT** 0xFc39e99e78524d7c891669F70d5aa92aB1041eb6 mcap $165k on $746/day, cheapest multiple on the chain at 0.6x but $19k of daily volume makes it untradeable. **bow.fun**, **Coinbarrel**, **Hookers** 0x833153ecb2c183702907e1142317a707137af954 (mcap $1.43M, NOT the same thing as HOOKR) all sub-$3k/day.
7. Picks and shovels: **INJOH 0x2cC0FAC44B8252f6B10208B091aFf2c94B4da77D** mcap $3.5M, burn 46,459,017 = 4.65% verified, takes 70% of INJOH trading fees from Pons plus 1% of every Sinjoh deployment. Levered to Pons without being a Pons competitor.

## CLAIM AUDIT
- "90% of PAIR protocol fees buy back and burn" — burn CONFIRMED (8.31% on chain), the 90% split UNRESOLVED, no doc found.
- "$400,000+ of PAIR burnt" (@pairdotfund, @cfm_sol) — TRUE at spot but MISLEADING. 83.1M tokens against ~$107k of implied lifetime protocol fees means the buybacks cost roughly $96k and are now marked at $445k. **Price every burn.**
- "$787M 24h volume on pair.fund" (@cfm_sol, 30,101 followers) — **FALSE, off by exactly 100x.** The project's own API says $7,872,323. The promoter misread his own source.
- "Noxa was the first independent launchpad, generated $13.4M-$14.5M in fees, beat Pumpfun in daily fees for 5 days, then stopped accepting launches mid-July" (@StarPlatinum_ 104k) — consistent with DefiLlama showing $4.22M of 30d fees and $0 revenue.
- PHISHING LIVE ON $PAIR: multiple accounts (@thenofirst, @emeline_lr, @FrancysFajardo4, @Asmaa91) posting "claim portal is open for $PAIR holders / check eligibility" with the CORRECT contract address and a link. Drainer pattern. Also a listing-vote bot farm ("Listing radar ping: $PAIR radar") and a reply-praise farm under @rizzkola, dozens of near-identical zero-like posts from 3k-30k accounts.
- Tier-one coverage: theunipcs (259k) is loudly long PONS and posts its revenue ramp; dontbuytops (14k) long PONS and CLAN; 0xkioto (12k) "pons is pve". **Zero tier-one coverage of any post-Pons pad, PAIR and o1 included.**

## LONG.XYZ LAUNCH RATE — measured 7 Sep 2026
- **LongLauncher contract 0x22e99278308b393ea1260859b181ad7e78f5eeed** (verified, name `LongLauncher`), event `LaunchCreated(address indexed poolOrHook, address indexed asset, address indexed numeraire, address poolInitializer, address launcher, bytes32 tickerKey, uint48 deployedAt, uint48 reservedUntil, string normalizedTicker)`, topic0 0xadc6f1f726f7c710f77ec06adc75f3bb964e5be19581b072c67f7b9b4039267b. Count these per day = launches per day.
- Doppler Airlock on RH: 0xeb7c034704ef8dcd2d32324c1545f62fb4ad0862. Pool initializer (single, all LONG launches): 0x4e3468951d49f2eea976ed0d6e75ffcb44a9a544. Token factory 0x1B37D3a72082029c44B35B604Ea473617580b69a, impl DopplerERC20V1.
- Daily LaunchCreated: 31 Aug 978 · 1 Sep 2,321 · 2 Sep 2,885 · 3 Sep 2,119 · 4 Sep 2,587 · 5 Sep 2,117 · 6 Sep 1,409. **Mean 2,059/day, rolling 24h 1,949.** long.xyz = 78% of all Airlock creates on the chain (2,646/day chain-wide).
- **GRADUATIONS ARE ZERO AND HAVE ALWAYS BEEN ZERO.** Three independent confirmations: (1) `Graduate(address)` topic 0xbd2bd570c963e5fe6bdc6422e5741c710099e75c6d44b6c73e6acc397429bdf7 on the initializer has never fired, lifetime scan blocks 0 to 57.01M; (2) long.xyz's own indexer returns 0 Assets with a non-null `asset_migration_timestamp`, all time; (3) the `GraduationPool` table holds exactly one row for chain 4663 and it is the 0xdead placeholder. Every asset is `asset_pool_type: "REHYPE"`, `asset_is_locked: true` — liquidity is seeded straight into a live v4 pool against the stock token, there is no bonding curve and no migration step. **Do not answer "how many graduate" on long.xyz as if it were a pump.fun-style pad; the concept does not exist there.**
- **long.xyz GraphQL API: `https://api.long.xyz/v1/graphql` (Hasura).** Cloudflare 403s it from the container; reachable from the Mac Chrome bridge with the page open on app.long.xyz. Chain id 4663 = Robinhood Chain. Tables: `Asset`, `Token`, `AuctionPool`, `GraduationPool`, `PoolVolumeHour`, `NumerairePrice`, `kalshi_market`, `builderCoins`, `productCoins`. **Row cap is 1,000 and there is no `Asset_aggregate`** — paginate with `asset_creation_timestamp:{_lt: cursor}`. It was intermittently failing on 7 Sep ("We're experiencing an issue with our data provider"); RPC log counting is the more reliable route.
- Robinhood Chain block time measured 0.1009 s = **856,472 blocks/day**. `eth_getLogs` caps at 10,000 results, so chunk at 100k blocks unfiltered or 200k+ with a topic filter. RPC needs a browser User-Agent.
- Dune was unusable on 7 Sep 2026: `{"error":"This api request would exceed your configured datapoint limit per billing cycle"}`. The monthly datapoint allowance is spent.

## THE DAILY SHORTLIST FOR LONG.XYZ — built 7 Sep 2026, replaces "sort by graduated"
- **`filterTokens` on api.long.xyz IS the Defined.fi / Codex filterTokens API, proxied.** Ben does not own a Codex key, and this gives him the feed for free. `filters:{network:[4663], tokenCreatedAt:{gte,lte}, ...}`, `rankings:[{attribute:X, direction:DESC}]`, `limit` max 100 per call. Filterable and rankable on: uniqueBuys/uniqueSells/buyCount/sellCount/txnCount/volume/buyVolume/sellVolume at 5m,1,4,12,24 · holders · notableHolderCount · marketCap · liquidity · change · **top10HoldersPercent · devHeldPercentage · sniperCount · bundlerCount · insiderCount · walletAgeAvg · swapPct1dOldWallet · isScam · trendingScore**. `filterPairs` is the same shape for pools.
- **CLOUDFLARE: api.long.xyz answers ONLY from inside a real browser.** 403 from the cloud container, 403 from the kookwatch droplet, and **403 from curl on Ben's own Mac** with correct UA/Origin/Referer. So any automation must run as JS in a Chrome tab sitting on app.long.xyz (Claude-in-Chrome `javascript_tool`, or a content script in the trenchbuy extension). Do not try to move it to the droplet, that has already been tested and failed.
- **Chrome `javascript_tool` times out at 45s.** Run multi-call sequences as a fire-and-forget async IIFE that parks its result on `window.__x`, then poll `window.__x` in a second call.
- **The Chrome tool redacts bare 0x addresses as "[BLOCKED: Sensitive key]".** Prefix them in the returned object (e.g. `'rh-'+addr`) and they come through intact. This is how to get full addresses out of any browser-side query.
- **The screen, derived from mechanism.** 2,000 launches/day, no bonding curve, so there is no "graduated" filter to lean on and 99% of launches never get a second buyer. The shortlist is a survivorship filter, not a quality filter. Hard gates: age 3h to 48h · holders >= 150 · top10HoldersPercent <= 35 · devHeldPercentage <= 2 · bundlerCount <= 3 · insiderCount <= 3 · liquidity >= $10k · buyVolume1 > sellVolume1. Rank by **holders-per-hour-alive x (1 + net1h buy dollars / sell dollars)**. Pull the universe as the union of the top 100 by `holders`, `uniqueBuys1`, `volume24` and `buyVolume1`.
- **Measured funnel, 7 Sep 2026: ~2,000 launched in the window -> 223 with any activity at all -> 16 passed the gates.** That 16 is the readable daily list.
- `top10HoldersPercent` is the highest-value single gate seen: it instantly kills things like a $29.2M "mcap" token with 37 holders and 99.8% in the top ten.
- **The trenchbuy deployer dot does NOT work on long.xyz coins.** `dep_table.json` keys on `contractFactory`, and every long.xyz token shares the one LongLauncher factory, so they all score identically. The per-creator signal here is the `launcher` address in the `LaunchCreated` event, which is a separate scorecard that has NOT been built yet. That is the next upgrade.

## LONGBOARD — the early-filter feed, built and live 7 Sep 2026
- [stated] Ben, 7 Sep 2026, correcting the first shortlist: SNOWBALL in it is a scam, Nasduck at $13.1M is "way too late", and **his goal is to catch them EARLY and to look at MORE projects, not fewer**. He can judge a coin himself in seconds (is the launchpad unique, is the stock-pairing joke good, does the infra look legit); what he cannot do is look at 900 a day. So he wants the dumb negative filters that take ~2,000/day down to 50-100, NOT a ranked "best coins" list. Scams, honeypots and serial-scammer deployers are what he wants stripped.
- **Service `longboard.service` on kookwatch, `/opt/longboard/longboard.py`, sqlite at `/opt/longboard/lb.db`, 5-minute loop.** Reads `LaunchCreated` from the RPC (no Cloudflare problem, unlike the long.xyz API), enriches every token aged 15min to 48h through DexScreener in batches of 30, writes `/opt/launchwatch/pub/longboard.{json,html}`. **Board: http://159.223.88.168/longboard.html** (sortable, auto-refreshes, full contract on every row, links to chart).
- **The five gates, all cheap and all negative:** >=15 buys in 24h · pool >= $5,000 · launcher made <= 20 coins in 8 days · marketcap/liquidity <= 25x · marketcap <= $1.5M. Tunable as constants at the top of the file.
- **Measured funnel, 24h to 7 Sep: 1,949 launches -> 384 that anyone traded at all -> 124 with >=20 buys -> 72 after the launcher and ratio gates.** Live board reads 104 of 401 while the launcher table is still filling; it tightens as the 8-day history accumulates. Both SNOWBALL and Nasduck are correctly excluded, SNOWBALL by the spam-launcher gate.
- **Contract-scam filtering is unnecessary on long.xyz and saying otherwise would be wrong.** Every asset is a minimal-proxy clone of `DopplerERC20V1` at `0x3Be8B97Fd0e713B5aBE0649Fa830223B6B4BC599`, identical bytecode every time, so there is no mint, no blacklist, no pause, no tax and no honeypot possible. The scam here is social: a spam launcher, a thin-curve fake market cap, insiders exiting. Hence the launcher-count and cap/liquidity gates carry the load that a bytecode scan would carry elsewhere.
- **`mcap/liquidity` is the fake-cap detector.** A REHYPE pool is seeded liquidity against a stock token, so a healthy fresh coin sits at 1.0x to 1.5x. SNOWBALL was 25x, Nasduck 30x. Anything above ~25x has a market cap that exists only on a thin curve.
- **DexScreener indexes only ~20% of long.xyz launches (384 of 1,949)**, and only once something trades. That is a free first filter but it is NOT the launch feed: for "see them all the second they launch" the `LaunchCreated` event is the only complete source.
- **launcher concentration, 8 days: 15,955 launches from 5,671 wallets, but 10 wallets made 100+ each** (top: `0x0e9f0e44de744d5ee86b8b8e3145d6901336d686` 1,180, `0x37379033390ae074a3689356eb394c7bda35c3e8` 806, `0xb8c70e8a52d215d1aeb8064821bbb8c97e2d0292` 702). Only 350 distinct launchers in any given 24h. The launch farms are a small, nameable set.
- **paste.rs is the file-transfer route into the droplet.** `curl --data-binary @file https://paste.rs` from the cloud container returns a URL the droplet can `curl` down. 0x0.st and transfer.sh are both blocked from the container. This avoids re-sending a whole script through a tool call.
- **The container cannot reach http://159.223.88.168 at all (503 from the egress proxy).** Verify droplet-served pages from the Mac or from the droplet itself, never from the container.

## METHOD NOTES
- Rank launchpads by `api.llama.fi/overview/fees/robinhood?dataType=dailyRevenue` filtered to category Launchpad. **DefiLlama's Launchpad TVL column is meaningless here** — it ranked NOXA #1 at $5.6M TVL when NOXA earns $0.
- A pad's `dailyFees` minus `dailyRevenue` is the creator share. Pons keeps 18.4%, o1 keeps 53.7%, Pools keeps 0%.
- `/summary/fees/{slug}` with `totalDataChartBreakdown` splits a multichain pad by chain. Essential: o1, Flap, Bags and Sentry all look bigger than their RH footprint until you split them.
- Burn check is one `balanceOf(0x...dead)` against `totalSupply()` on the RH RPC (browser User-Agent required). Do it on every pad token; it separates PONS (29.32%), INJOH (4.65%) and PAIR (8.31%) from POOLS (0.00%) and STONKBROKER (0.05%) in one call each.

## GRADUATIONS PER DAY — measured 13 Sep 2026 (Ben's auto-buy agent question)
- **PONS V2 IS THE ONLY PAD ON THE CHAIN WITH A REAL GRADUATION EVENT.** Factory `PonsV2LaunchFactory` **0x7ed598bcef8bd9edd8c97a195c6d13f40801ec7e** (reached via deployer `PonsV2LaunchDeployer` 0x3711ceA4feaDE896C913C68F01Eda97Cb06D1A42 -> `factory()`). Event `PoolGraduated(address indexed token,uint256 positionId,uint256 tokenAmount,uint256 pairTokenAmount)`, topic0 **0x0a44ef75df69c534f43cd6c1aa3ef8983065fe5fe79ef9e79f6494e6f258c259**. Also `TokenLaunched(...)` topic0 0x8d4aad4953d0ca700d468f3753aa14432d1b35b43ec6409f051fb6aa43a89607.
- Daily graduations (UTC): 6 Sep 545 · 7 Sep 480 · 8 Sep 495 · 9 Sep 369 · 10 Sep 306 · 11 Sep 373 · 12 Sep 297. **7d total 2,865, mean 409/day.**
- Pons V2 launches 12 Sep = **16,553**, graduations 297 = **1.79% graduation rate**. A graduation is already a 1-in-56 filter.
- **Pons V1 `PonsLaunchFactory` 0xa5aab3f0c6eeadf30ef1d3eb997108e976351feb has NO graduation event.** Its `TokenLaunched` carries a `pool` and `dexFactory` — the pool is seeded at launch, straight into Uniswap V3. Same shape as long.xyz. So "graduations on Pons V1" is zero by construction, not by inactivity.
- Launchpad revenue 24h on 13 Sep: Pons V2 $866,423 of $934,129 chain-wide = **92.8%**. Pons V1 $28,813 · StonkBrokers $13,812 · o1 $11,381 · Flap $5,872 · Cook Market $2,105 · LetsCash $1,599 · PAIR $1,421 · Pez $1,071 · Bags $814 · AGNT $755 · Hookers $508. Non-Pons pads were not traced to contracts; their revenue share bounds their graduation count.
- Chain-wide Uniswap V3 `PoolCreated` runs ~485/day but is NOT graduations: ~41% are manual LPs through the NFPM 0x73991a25c818bf1f1128deaab1492d45638de0d3 and ~14% through direct-launch factories like `RobinhoodDirectV3MetadataLaunchFactoryV1880` 0xfb21934bb01b4d7b83beb8af6e6fd553f049e632. Do not use pool creation as a graduation proxy.

## SOCIALS ARE ON-CHAIN ON PONS V2 — do not use an indexer for this
- Every Pons V2 token is a `PonsV2LauncherToken` with **`socials()` selector 0x53cd512a** returning five strings (twitter, telegram, discord, website, farcaster) and **`getTokenInfo()` selector 0xabb1dc44** returning deployer, logo, description and the same socials struct. Free, no key, readable the instant `PoolGraduated` fires. 2,865 tokens read in ~90s in eth_call batches of 40.
- **DexScreener is NOT usable as the socials source: it had already dropped 767 of the 2,865 graduates (27%) entirely** (`pairs:null`), and it only indexes a token once it trades.
- Breakdown of the 2,865 graduates: twitter 2,551 (89.0%) · website field 1,476 (51.5%) · telegram 179 (6.2%) · discord 9 (0.3%) · nothing 246 (8.6%). Combinations: both 1,408 (49.1%) · twitter only 1,143 (39.9%) · website only 64 (2.2%) · neither 250 (8.7%).
- **HALF THE "WEBSITES" ARE NOT WEBSITES.** 926 unique domains across 1,476 fields. Top entries: x.com 99, ponsfamily.com 91 (the pad itself), mademen.family 42 (one farm), github.com 31, otcdesks.cash 30 (one farm), reddit 25, robinhood.com 19, stonkfun.xyz 19, crumbs.family 17, truthsocial 16, youtube 15+12, instagram 15, huggingface 11. Blocklisting pad/social/media/aggregator domains takes the website gate from 51.5% to **37.6%**.
- Copy rates inside one 7-day window: **37.3% of graduates share a ticker with another graduate**, and **31.9% share a Twitter handle with another graduate**. A rolling handle/domain/ticker table is the single highest-value filter and no third party supplies it for this chain.
- **The funnel for Ben's agent, per day: 409 graduate -> 364 have twitter -> 211 any website -> 154 real website -> 148 real website + twitter -> 108 original socials (handle and domain unique) -> 87 also a non-duplicate ticker.**

## GMGN ON ROBINHOOD CHAIN — 13 Sep 2026, Ben asked about using it
- GMGN DOES support the chain: `gmgn.ai/?chain=robinhood` renders a HOOD Trenches board (New / Almost bonded / **Migrated**), and token pages work at `gmgn.ai/robinhood/token/{address}`.
- **But its API coverage for this chain is thin.** `GET /api/v1/token_info/robinhood/{address}` = 200 and returns only address, symbol, name, decimals, logo, biggest_pool_address, open_timestamp, holder_count, circulating_supply, total_supply, max_supply, liquidity, creation_timestamp — **zero social fields**. `/defi/quotation/v1/rank/robinhood/pump/1h` answers `{"code":40000400,"msg":"network not supported"}`. Chain slug is `robinhood`; hood/rhc/rh/4663 all return invalid argument. The Trenches board is fed by websocket, not by a REST list endpoint I could find.
- **gmgn.ai Cloudflare-403s the Cowork container on every path including the homepage.** Only reachable through Chrome on Ben's Mac. Same class as the fomo/long.xyz walls.
- The "Original Socials / Exclude Honeypot / Exclude Non-Renounced / LP Burnt" filter panel Ben screenshotted is the Solana side. **The Token Audit boxes are meaningless on Pons V2 anyway:** every token is identical `PonsV2LauncherToken` bytecode, so honeypot, mint, blacklist and non-renounced cannot occur. Original Socials is the only box worth having and it is reproducible from `socials()` plus a rolling seen-table.