product updates: march 19, 2026

40+ commits since the last update. new agent, completely overhauled trading controls, smart position sizing, strategy refocusing, wallet fixes, and infrastructure hardening. this is a big one.

but before the feature list, some context on where parasol is going — because the direction explains the decisions.

---

why parasol is becoming more interactive

when we launched, parasol was closer to a trading bot. set your risk profile, fund the wallet, let the agent run. that model works for a specific kind of trader — someone who wants fully autonomous execution and doesn't need to touch anything.

but watching how people actually use parasol told a different story.

most traders don't want to hand over full control. they want to see what the agent is doing. they want to step in when they have conviction. they want to sell half a position when their gut says it's time, buy more of a token they believe in, and override the agent's sizing when they have an edge. they want the agent to handle the parts of trading they're bad at — discipline, speed, watching charts at 3am — while keeping control over the parts they're good at.

that's not a bot. that's an assistant.

the updates in this cycle are all pushing in that direction. every new control — quick-buy from the watchlist, 2x buy on positions, inline sell buttons, custom amounts — is about reducing the friction between "i want to do something" and "it's done." parasol finds the tokens, filters the garbage, monitors the positions, manages the exits. you step in when you want to, and the rest runs itself.

the three modes from our trading assistant article are the framework:

the updates below make modes 2 and 3 dramatically faster and more intuitive. mode 1 also benefits from smarter sizing, better balance tracking, and a new agent that adds social intelligence to the scan.

every feature ships with the same principle: parasol should never be in your way when you want to act, and should never need your attention when you don't.

---

grok oracle agent: X-powered token discovery

a new agent template: Grok Oracle. while sienna scans charts and claudia monitors pump.fun, grok scans conversations.

grok pulls token mentions from X in real time, resolves contract addresses via DexScreener, and feeds them into the standard parasol analysis pipeline — the same manipulation filters, scoring, and risk management that every other agent uses.

the thesis: tokens that are trending in conversations before they trend in charts are where the edge is. by the time a token shows a volume spike on DexScreener, the first movers found it on X. grok tries to be one of those first movers.

grok complements sienna (technical analysis) and claudia (pump.fun early momentum) by adding a social signal dimension to the agent lineup.

also in this cycle: the access code gate has been removed. new users can start immediately.

---

quick trading controls: act from any view

this is the biggest UX overhaul since launch. the goal: every token the agent shows you should be one click away from a trade.

watchlist quick-buy

the watchlist was read-only. it showed you what the agent was considering — confidence scores, reasons, trade flow — but if you wanted to buy, you had to click the token, open the chart, find the buy panel, and enter an amount. four steps between conviction and action.

now: one green BUY button on every watchlist item. one click, default amount, done. the button shows "buying..." while it processes.

this turns the watchlist from a research view into a trading surface. see a token with 85% confidence and strong buy pressure? act immediately.

2x buy on open positions and wallet holdings

every position row — both agent-managed positions and external wallet holdings — now has a green BUY 2x button. it buys 2x the current value of that position in SOL.

the use case: you're up on a token and want to double down. previously this required opening the chart, calculating the amount, and manually entering it. now: one click.

consistent sell controls everywhere

wallet holdings previously had a different sell interface than open positions — a percentage input box with presets. that inconsistency is gone. both now use the same button row:

BUY 2x (green) | 50% (red) | SELL X.XXX (red) | ✕

same buttons, same layout, same behavior. no context switching between different parts of the dashboard.

sell button overhaul

the sell area was completely restyled earlier in this cycle:

these controls matter because parasol is a portfolio management tool as much as a scanner. once you have positions, you need fast, intuitive controls to manage them. every extra click or unclear button is a missed opportunity — or worse, a late exit.

---

smart position sizing: wallet-balance-aware

position sizing was a fixed USD amount. a wallet with 0.5 SOL and a wallet with 50 SOL would attempt the same $12 trade. that's wrong.

sizing is now derived from your wallet balance:

manual buys bypass the per-trade cap entirely. if you choose a custom SOL amount, parasol honours it without interference. the sizing logic only governs autonomous trades.

this is core to the portfolio management angle: intelligent capital allocation matters more than any individual entry.

---

claudia refocused: early momentum specialist

claudia's strategy has been tightened to focus on what she does best — catching tokens early in their lifecycle.

post-peak rejection — claudia now passes on tokens that have already had their big move. if the price is significantly below its recent high, the momentum she's looking for has already played out.

first-burst detector — a new signal that boosts confidence for tokens showing concentrated buy pressure in their first 90 seconds of trading. early accumulation is exactly the pattern claudia is designed to catch.

wider bonding phase coverage — lower liquidity floor, near-graduation tokens allowed, smaller market caps eligible. claudia's edge is catching tokens before they're on anyone's radar, so the universe needs to be wide even as the quality gate is strict.

no cycle cap — if 5 tokens pass the filters in one scan, claudia trades all 5. the post-peak filter and tighter entry gates handle quality control.

---

fee split: agent token buyback

platform fees are now split 50/50:

implemented via pump.fun's agent payments SDK. fee splitting happens at execution time.

---

priority fee upgrade

transactions were landing late or failing entirely during high-congestion periods.

slightly higher cost per trade, dramatically better fill rates on tokens that move quickly.

---

wallet and balance fixes

five interconnected bugs were causing inaccurate balances and false trade rejections. this was the most common source of user complaints and all five are now resolved.

1. wallet balance display

the profile wallets section showed incorrect SOL balances. the Helius API call excluded native SOL from its response, so the displayed balance was derived from token USD values divided by SOL price — a completely wrong number.

fixed: native SOL is now returned directly from the Helius API.

2. live wallet not tracking active wallet

the "LIVE WALLET" section always showed the primary Turnkey wallet's balance, even when an imported wallet was set as active.

fixed: the live wallet section now reactively tracks whichever wallet is currently active.

3. agent balance double-counting

after fix #1, the agent dashboard showed double the real balance. SOL was being counted from two separate API calls and summed.

fixed: the dashboard now uses token-only values from the holdings API, so native SOL is counted exactly once.

4. insufficient balance rejections

the portfolio's internal balance ledger was only set at agent start and never updated. external deposits or withdrawals made it stale, causing trades to be rejected with "insufficient balance" when the wallet had plenty of SOL.

fixed: the portfolio syncs with the real wallet balance at the start of every trading cycle.

5. budget cap re-triggering

the budget tracked gross outflow only — buys incremented it, sells never credited it back. after a few buy-sell cycles, the budget hit its cap even though capital had been recycled.

fixed: sells now credit the budget. it tracks net exposure, not gross spend.

these five fixes are deeply connected — inaccurate balances caused false rejections, which caused users to hit the budget cap early, which compounded the frustration. fixing them as a group eliminated the entire class of problems.

---

leaderboard fixes

live trades weren't being recorded to the points system on close, showing 0 trades for active users. a secondary bug caused some trades to be counted twice.

fixed: every position close now records win/loss to the leaderboard. deduplication prevents double-counting.

---

chart and UI improvements

chart auto-scroll — the chart no longer snaps to the right edge on every new candle. if you've scrolled to examine history, it stays where you put it.

chart fragmentation — WebSocket ticks were creating micro-candles. ticks are now aggregated before rendering.

chart header — token logo increased from 20px to 28px with hover-to-enlarge. token name larger and more legible.

dashboard readability — comprehensive pass across the agent dashboard: larger fonts, better contrast, consistent panel styling.

these matter because parasol doubles as a research and monitoring tool. if you're using the scanner and watchlist to research tokens before trading them, the charts and data need to be easy to read at a glance. readability is functionality.

---

watchlist upgrades

beyond quick-buy, the watchlist gained:

the watchlist is becoming the command center for mode 2 trading: curate what you want to watch, act when something moves, let the agent handle the rest.

---

multi-wallet position tracking

positions are now stamped with the wallet address that opened them. sells route through the correct wallet automatically, even if you've switched active wallets since.

this is essential for traders managing multiple strategies across different wallets — a common pattern for separating risk profiles or tracking performance per wallet.

---

infrastructure

---

the direction

every update in this cycle serves the same thesis: parasol is most useful when it handles everything you don't want to handle, and gets out of the way for everything you do.

that means:

parasol isn't one thing. it's a system that works at whatever level of involvement you choose.

---

summary

40+ commits across two days. the trading controls let you act instantly from any view. the balance and budget fixes eliminate the most common complaints. the strategy improvements make the agents sharper. and the direction is clear: a platform that works as a research tool, a trading assistant, a portfolio manager, and a full automation engine — depending on what you need.

stay sharp. parasol keeps you covered.