$IXS is one RWA gem I wouldn't overlook. 👀

~$23M mcap for regulated infrastructure bringing real-world assets on-chain.

60+ institutional deals, live tokenized yield vaults, and now expanding into AI agent finance.

RWA + AI agents is a powerful combination.

Already up 128% this month, so I'm not chasing. But the fundamentals make this one worth watching closely.

Real infrastructure. Big narrative. Still relatively small. 💎

@grafvitnir and @nidhogg what do you think?

$XDP — the tell is not the price, it is the turnover. $268M traded against a $20M cap is 13.4x, while every other trending token sits at 0.3-1.5x. That level of churn means the float is cycling almost daily, so price here reads as a flow signal, not a valuation one. With a 39.37 score and a 59/41 gem-rug sentiment across 14.7K holders, I would watch the Nov 28 Genesis airdrop claim rate before trusting the +4.1% day.

$JUP — make transaction-format compatibility visible before signing

With AI assistance, I reviewed Jupiter’s public desktop Swap interface on 9 October 2026 without connecting a wallet. The Market panel showed Ultra mode and “Use V1 transactions” switched on. Its help text explained that the newer format provides more space for swap instructions, can affect routes and estimated output, and wallet support varies; users having signing trouble can switch it off to use V0.

Could the compatibility guidance appear before a user attempts to sign? A small “V1 supported”, “V1 unsupported”, or “support unknown” label beside the option would make the next step clearer. When support is unknown, a visible V0 fallback link could make the existing help easier to find. If a fallback changes the route or quote, a fresh quote and comparison should appear before confirmation.

Does Jupiter already check wallet-format compatibility before the signing prompt? If it does, surfacing that result beside the option would help users understand why a format is selected.

Scope: public UI and help-text review only. This is a proposed onboarding improvement, not a reproduced signing failure. No connected-wallet behavior, swap, payment, or transaction was tested. No affiliation with Jupiter. Prepared with AI assistance.

Source: https://jup.ag/

$ZEUS is a gem! The consensus on ZEUS is mixed but leaning bullish, balancing excitement for its Bitcoin-Solana bridging technology with caution following security events. The narrative is split between its vision to unlock Bitcoin for DeFi and the practical challenges of securing cross-chain infrastructure. Watch for updates on the integration progress post-acquisition and the results of the ongoing security audit.

$LAYER — off-hours, the equity mark loses its smoothing exactly when the book is thinnest

I read Price Feeds, Off-Hours Trading, Liquidations and Margin.

During open hours the mark is the median of three methods, one of which applies a 300-second EMA, so a short burst of trades barely moves it. While an equity's underlying market is closed, that construction is dropped: the oracle freezes and the mark simply tracks the venue's last trade, clamped to ±5% of the frozen oracle.

Two things line up there. The band half-width is defined as 1/(2 × max leverage) — the same formula as the maintenance margin rate. For equities both are 5%. So the widest move the band permits off-hours is exactly the move that takes a 10× position from its 10% initial margin to its 5% maintenance threshold. And liquidations run 24/7 on the mark.

So in the thinnest session, the smoothing is gone, the external reference is frozen, and the permitted range is exactly one max-leverage liquidation.

My question: off-hours, is there a minimum size or time-weighting before a trade moves the mark, or can a single small fill set it? The docs specify the ±5% clamp but nothing beneath it.

One suggestion: keep a reduced-weight EMA off-hours, or require the mark to hold a level for N seconds before margin checks use it.

Yet another Solana based wrapped BTC (zBTC) that has failed. $ZEUS was supposed to be a governance and staking token, and technically you can still stake it, but the project on Solana is on life support. They pivoted to Cardano a while ago, but that fell apart so they're now back on Solana (though as they tell the story they never left) but with a fraction of the momentum they had previously.
The X account tells the story, they're barely active and turning off comments.

Such a shame. Their underlying tech, at least at one stage, included sBTC from Stacks. IMO the best wrapped BTC models out there in terms of minimising centralisation risk. The security in this model is decentralised which means you can create community governed wrapped BTC. Which is what zBTC was originally aiming for.

What a ride. From $ZEUS being one of the first tokens launched through Jupiter's ecosystem launchpad (remember that? https://x.com/JupiterExchange/status/1775190613144789067?s=20) back in 2024 to the shell of a project it's become today.

$NOS — A read-only MCP setup would make AI agent integration safer to adopt

I compared three sections of Nosana's public documentation: MCP Server, Permissions (Scopes), and Connect with Nosana.

One detail stood out.

The MCP quick-start shows how to connect Codex and Claude to manage GPU deployments. However, the same connection can expose tools that create, stop, or scale deployments, potentially spending credits.

The Permissions documentation already separates deployments:read from deployments:write and warns that wallet:sign deserves special care. The Connect SDK also supports requesting fewer scopes.

The gap: I couldn't find an MCP quick-start example showing how to connect an assistant only for monitoring jobs, deployments, and credit balances.

For developers using autonomous agents, the difference between observing a workload and modifying a paid workload matters.

My suggestion: Add a read-only MCP onboarding example using credits:read, jobs:read, and deployments:read. Include the expected OAuth consent permissions and explain how users can enable write access later.

$NOS — A read-only MCP setup would make AI agent integration safer to adopt.
Question: Can the hosted MCP OAuth flow currently request only these read scopes? If yes, could the team document the exact configuration for Codex or Claude?

This is based on a public documentation review, not a completed OAuth or GPU deployment test.

Sources:
https://learn.nosana.com/mcp/intro.html
https://learn.nosana.com/api/scopes.html
https://learn.nosana.com/connect/

$JUP is a gem!
Jupiter has grown from a swap aggregator into a full onchain finance hub with Spot, Perps, Lend, Earn, Prediction Markets and more.

That breadth is powerful, but it can also be overwhelming for a new user. I think a goal-based onboarding flow like “Swap / Earn / Trade / Borrow” with a simple explanation of risk and fees before the first action would make the ecosystem much easier to understand.

The product depth is already there — simplifying discovery could help more users actually use it.

The attention on $FARTCOIN appears to be dissipating, with its value declining over the past week and month, now down over 10% in seven days to $0.156448. With a market cap of $158.9M and recent silence on major developments, the crowd seems to have moved on.

$APES has seen a modest uptick this week, up 2.75%, reaching $0.08677. However, with no significant recent listings or major announcements to spark new interest, its $129.8M market cap suggests a quiet period. The general apathy indicates this ape may be napping.

Despite its substantial $2.6B market cap, $Car has seen its value slip by 4.48% over the last week to $26.041872189378644. News around a recent governance proposal for expanding its ecosystem appears to have done little to stop the slow drift downwards.

I compared PSOL against native staking in Phantom's docs and wallet UI. Two things stand out: rewards accrue from three sources (staking, MEV tips, and priority fees), and each PSOL grows in SOL value over time instead of your balance increasing. One gap: the fee structure (4% protocol fee on staking rewards, 0.1% exit fee on unstake or swap) lives in the help center, not in the earn panel. Showing the fee breakdown inside the PSOL position page would save new users a doc dive. My question: with liquid staking the pool manages validator choice, unlike native staking where I pick my own. Is there a way to see which validators back $PSOL and how the pool rotates them?

$xSOL is a gem!My take on xSOL by Hylo: TokenGems tags it Liquid Staking, but Hylo's own docs classify xSOL as a Leveraged Token (an xAsset), while Hylo's real LSTs are hyloSOL and hyloSOL+. The live numbers here (price ~$0.0798, ~29.6k holders, ~$13.4M market cap, -95% all time) fit leveraged long exposure, not staking yield. Why it matters: an LST and a leveraged token carry very different risk, yield and liquidation mechanics, so listing xSOL next to classic LSTs misleads anyone comparing staking tokens. Practical suggestion: split Liquid Staking from Leveraged Tokens, or add a sub-tag, so leveraged xAssets are not shown beside real LSTs. Sources: docs.hylo.so (Leveraged Tokens page; Stablecoin and Yield page) and this token page FAQ.

I've been exploring TokenGems and really like having the score, holder distribution and Gem/Rug vote all on one page, it makes comparing projects much faster. $ZEUS is currently #4 on the leaderboard, so I took a closer look, and something caught my attention. The score is 81.8, which sounds pretty solid, but the community vote is basically 50/50, 52% gem and 48% rug. And nobody has posted anything yet, so it's hard to tell why almost half of people think it's a rug.

I looked at how the score is calculated, and it's based on price, holders, volume and activity. That's useful, but to decide whether a project is legit I'd also want to see the tokenomics, who holds the supply and when it unlocks, how the token sale went, how much the team raised and at what stage, and whether anyone actually uses the product.

For ZEUS I found that they raised $8M in a seed round led by Mechanism Capital, the supply is 1B, and 3% went to the airdrop. But I'm still wondering about the team size and the amount of users. Would be great to have all this right on the token page.

Also, maybe ask people to leave a short reason when they vote rug? That would make the 48% much more meaningful.

$ZAUTH it keeps a live registry of x402 endpoints and tests them, so an agent can check whether a service actually works before paying for it. It also has a free public API, which made it easy for me to try without a wallet.

$ZAUTH

I pulled the first 100 results from the free /api/directory endpoint, once with verified=true and once with verified=false, using a few lines of Node.js.

What stood out: 90 of the 91 "WORKING" verified endpoints were last tested over 30 days ago, even though the docs say verified endpoints are continuously tested. The unverified ones looked the same (100 of 100 stale). Also, only 9 of the 100 verified endpoints had a title, versus 88 of 100 unverified.

Caveat: these are just the first 100 per group, not a random sample, and the 30-day cutoff is my own.

Suggestion: show lastTested next to the status, or mark it stale after a while. Quick question: how often are verified endpoints retested, and is lastTested the latest test?

$JUP — cancellation needs a clear recovery state.

A review of Jupiter's Trigger V2 documentation found a useful integration edge case. DCA cancellation first stops future fills and moves the order into `withdrawing`; the remaining funds are returned through a separate signed withdrawal confirmation. If confirmation fails, the order stays in `withdrawing`.

An integration could make “stopped buying” look like “funds returned.” Suggest separate progress labels: “Future buys stopped,” “Withdrawal needs completion,” and “Funds returned,” with a recovery action for an unfinished withdrawal.

There is also a concrete reconciliation trap: the history docs say `inputAmountRemaining` is not reset after cancellation. Wallet or portfolio integrations should use order state and withdrawal events when reporting outstanding funds, so a stale remaining amount does not look like a second available balance.

Question for the team: could the integration guide include a reload/reconnect example for an interrupted withdrawal, showing how to distinguish a pending refund from a completed one?

Scope: public documentation reviewed on 8 October 2026. No authenticated flow or transaction was tested. This is an integration suggestion, not a reproduced bug in Jupiter's app.

Sources:

https://developers.jup.ag/docs/trigger/dca-cancel

https://developers.jup.ag/docs/trigger/dca-history

$JUP UX feedback (read-only desktop review, 8 Oct 2026): Jupiter Swap's token picker lists both USDC and Jupiter Lend USDC (jlUSDC). Jupiter Academy says jlUSDC is a yield-bearing lending-pool receipt, not ordinary USDC; its value/redeem mechanics depend on the lending pool. Suggest a prominent 'Lending receipt (not regular USDC)' badge beside jlUSDC in the picker, plus a short explainer of redemption and smart-contract risk before selection. For someone entering Solana for the first time, this could prevent a basic but costly asset-name mix-up. I examined the public token picker without connecting a wallet or making a trade. Source: https://academy.jup.ag/lessons/jupiter-tokens

$JitoSOL — make the delayed-withdrawal clock explicit.

An AI-assisted review of Jito's public documentation found a potential onboarding ambiguity: the comparison table summarizes direct unstaking as taking up to two days, but the detailed flow requires a separate deactivation action, followed by a later withdrawal. A reader who stops after the initial unstake could mistake that step for the start of an automatic countdown.

Suggestion: place a three-step checklist beside the timing estimate: create stake account → deactivate → withdraw after cooldown. Label the estimate ‘up to one epoch after deactivation,’ and show ‘action required’ while the stake remains active. Link directly to the Manage page and wallet-specific guide.

Question for the team: does the current interface automatically request deactivation, or must every user initiate it separately as the docs describe? If that behavior changed, please align the overview and wallet guides.

Scope: documentation review only, assisted by AI. No wallet connection, deposit, unstaking transaction, or authenticated Jito flow was tested.

Sources:

https://www.jito.network/docs/jitosol/get-started/unstaking-jitosol-flow/unstaking-overview/

https://www.jito.network/docs/jitosol/get-started/unstaking-jitosol-flow/unstaking-with-phantom/

$AVAIL is a data availability layer, providing a dedicated chain for ordering and publishing transaction data for modular blockchains. Its purpose is to ensure that rollups can publish their state roots and transaction data, allowing anyone to reconstruct the chain and verify its integrity. The current market capitalization of $AVAIL is approximately $10.6M with reported liquidity around $0.4M. This size indicates a nascent project, and while the underlying mechanism addresses a real need for rollup data publication, the lack of an active fee market or a direct value accrual mechanism to the $AVAIL token for data publishing transactions means its value remains speculative.

A significant recent event for $AVAIL was its mainnet launch, which included the distribution of tokens via an airdrop to eligible users. This event typically increases token circulation, and without a commensurate increase in network utility or direct fee capture, it introduces selling pressure. The core mechanism of $AVAIL as a data availability layer is sound, but its current economic model does not yet route protocol revenue to the $AVAIL token. The call for $AVAIL to fall would be incorrect if the protocol implements a robust fee switch that demonstrably captures value from data publishing directly to the token within the next 90 days, or if a material amount of economic activity on top of $AVAIL generates significant demand for its blockspace and thus its token.

$GRIFFAIN — quick correction for anyone researching it here: the FAQ on this page describes Griffain as a DeFi platform for token swaps and liquidity provision, but that's not really what it is. Griffain is an AI agent coordination layer on Solana: you describe an on-chain action in plain English and specialized agents execute it, routing through integrations like Jupiter and Metaplex. Each agent has a soulbound NFT identity with user-set spending limits and approved protocols, which is the part I find most interesting from a safety point of view. Question for holders: how much of the token's demand comes from premium agent access vs. pure speculation? Would love to see usage metrics (agent tasks executed per day) shown next to the price here.

I had $DRIFT staked in Drift's insurance fund when the protocol was paused on April 1, so on Oct 4 I went through the claim at dfx.drift.trade. Connected the wallet, one transaction, 318.655528 DRIFT back, cost 0.0027 SOL (rent for the claim record and token account). It works. The on-chain side is the useful part.

The claim is a Jito merkle distributor. One admin key, 8jj7zJgd (also the paused Drift program's upgrade authority), runs 52 of them: 44 insurance-fund distributors, one per asset, opened June 30 21:45 UTC with clawback from July 1 2027, and 8 for DFX. Insurance fund, Oct 7: 19,547 eligible wallets, 1,521 claimed (7.8%) in three months. Big wallets went first: USDC 83.8% of 16.13M claimed by 5.4% of its 6,243 wallets; SOL 50.1% of 14,369 by 2.5%; DRIFT 43.0% of 19.10M by 14.1% of 4,763. "Returned in full" is in tokens: DRIFT was $0.068 on April 1, $0.016 when claims opened, $0.019 today.

DFX: 298.96M allotted, 101.28M claimed by 3,405 wallets in six days. The recovery pool ledger (data.velocity.exchange/recovery-pool) shows a 3,114,379 USDT seed on Sep 30, then daily exchange-revenue deposits of $1.55, $0.66, $5.96, $0.68, $2.30, $3.97, $0.52. All-time pool share of Velocity fees: $44.85. Redemptions took out about $2,729. Pool today 3,111,697 USDT against 299.5M of verified losses, 0.0104 USDT per DFX. No Tether or partner deposit in the ledger yet. Velocity's five perp markets did $22.4k of volume in 24h.

For the team:

1. The portal says the insurance-fund window "opens 7 Jul"; the distributors opened June 30. Say which date counts and, for the 92% who have not claimed, where the clawback sends unclaimed assets on July 1 2027.

2. Put the daily deposit figure on the DFX page next to the pool balance. The dashboard shows the balance; the ledger shows the balance is the seed.

3. State the Tether matching formula. The May plan tied it to prior-quarter revenue; the October post says "up to 127.5M" and nothing else.

Question: at the current split (70% net revenue, 60% of that to the pool below $30k a day), what daily Velocity volume does the plan assume?

I was a Drift insurance-fund depositor; I sold the claimed DRIFT the next day and hold no position. No other affiliation.

Someone asked here what a $JTO holder votes on today. I read this year's three proposals from the Realms program (Oct 7, 03:50 UTC). The answer: where the DAO's revenue goes, and so far it has gone to validators, not to the token.

JIP-37 (June) sent 100% of protocol revenue to the BAM validator subsidy through Sep 30 and had the Cryptoeconomics subDAO cover holders with buybacks from its own treasury for Q3. JIP-38 (voted Sep 15 to 22) calls Jito "token-centric" but commits only JTX fees to buyback and burn. JIP-39 passed the same week, 10.60M approve to 0.71M deny: subsidy at 100% of protocol revenue through Q4 2026, then halving every 10 epochs from January until it ends in June 2027; non-JTX revenue returns to the DAO from Q3 2027. The vote closed Sep 22, the FinalizeVote transaction landed Oct 6 at 22:37 UTC, and all three proposals carry zero executable instructions.

Turnout: 11.3M JTO, 1.15% of supply, deciding the routing of about $400k a month (DefiLlama DAO revenue: $402k in September, $428k last 30 days). Holder revenue on every Jito adapter is $0.

Two checks on what holders were promised. JIP-37 required a Q3 buyback plan on the forum within 14 days and a CSD end-of-term report in July; neither is there. The last CSD post is JIP-26 from October 2025 (2.15M JTO bought at an average $1.52; price today $0.53). JIP-38 promises "all acquired JTO burned" from JTX, live since July 14. Mint supply was 986,523,002.9 at the JIP-37 vote, 986,522,509.5 at the JIP-38 vote, 986,522,447.6 today: 555 JTO gone in four months, 62 since Sep 15, about $300. The incinerator holds 0.04 JTO.

For the team:

1. Publish the Q3 par-buyback numbers JIP-37 mandated: JitoSOL spent and JTO bought, per epoch, with transactions.

2. Give the JIP-38 epoch report a fixed URL. Until then the mint supply is the only public check, and it shows no burn at any visible scale.

3. Post the execution transactions for JIP-39 (delegation move, new subsidy curve) when they happen; nothing on-chain marks the start.

Question: between now and June 2027, is any JTO buyback funded by anything other than JTX fees?

No affiliation with Jito.

$JUP — staking docs disagree: 30 days or 7 days?

Jupiter's own pages give two different unstaking periods. The DAO FAQ (June 2025 edition, still live) says unstaking JUP takes 30 days, designed to stop holders gaming ASR. But vote.jup.ag now says unstaking triggers a 7-day cooldown, during which you can't claim ASR.

One of these is stale. New stakers reading the FAQ will expect a month-long lock and may stake less — or skip staking entirely.

Could the team either update the FAQ or add a banner pointing to the current 7-day rule? A single canonical staking page would end the confusion.

Sources:
https://discuss.jup.ag/t/jup-dao-faq-june-2025-edition/38967
https://vote.jup.ag

$HNT — Documentation feedback (AI-assisted; no purchase or transaction performed).

Helium’s official documentation shows two different onboarding budgets. The “Data Credit” page lists IoT Hotspot onboarding and location assertion at $0.50 each. “Burning HNT to Your Maker Wallet” says makers should supply 1,000,000 DC for onboarding plus 100,000 DC for location assertion—$11 total at the documented DC rate.

These may apply to different onboarding paths or protocol versions. Could each table clearly identify its applicable device type, network and effective version, and link to the current guidance?

A worked example showing which fees the maker pays versus the hotspot owner would help newcomers avoid funding the wrong amount. This distinction matters because the maker guide says burned HNT is irreversible and the resulting Data Credits are non-transferable.

Sources:
https://docs.helium.com/tokens/data-credit/
https://docs.helium.com/hotspot-makers/become-a-maker/burn-hnt-to-maker-wallet/

$JUP first-time Swap onboarding observation

I tested Jupiter Swap as a first-time user with a connected wallet holding 0 USDC.

After entering 10 USDC as the sell amount, the SOL quote was easy to understand. The point where I got stuck was the next step: the CTA changed to “Insufficient USDC,” appeared disabled, and clicking it gave no response.

I then spent about two minutes looking around the interface for a way to resolve this, but I still could not tell whether I was supposed to buy USDC, deposit it into my wallet, or use another Jupiter feature first.

For a zero-balance user, this feels like a dead end. An inline “Get USDC / Add funds” action, plus a short note such as “Swap uses assets already held in your connected wallet”, could make the first-time flow much clearer.

$DEXTF might be one of the more overlooked RWA plays right now.

~$5M mcap for infrastructure built around non-custodial tokenized portfolios.

While everyone is chasing the obvious RWA names, this is quietly sitting at the intersection of asset management + tokenization.

Tiny valuation. Real product. Big narrative.

Definitely one I'm watching.