nothing is financial advice. think for yourself. don’t be a dork. sibyl to a bibyl
Joined August 2017
- Tweets27.4K
- Following6.7K
- Followers2.8K
- Likes60.7K
kudeta retweeted
The Sibyl Memory product now is officially live and available for purchase.
Let us (or me) know in case you want to move to superior agentic memory.
Why Sibyl Memory Pro?🧠
Thousands of user hours. Proven effectiveness.
Self-Learning: True experiential growth and emergent behaviors come from teaching an agent things it does not already know. With Sibyl, you can build agents that learn the unique parts of your work, so you save money on all future work.
Private by Default: Your memories, your device, your decisions. Sibyl Labs never obtains your personal data other than what is necessary for billing and analytics.
Auto-Debloat: Automatically enforced environment hygiene that cleans up your environment, reduces active tokens in context, and compounds your savings over time.
Learn more: sibyllabs.org
kudeta retweeted
we pointed our new larp check at our own coin first. 7 tools answered on the mcp server, 2 outside wallets paid for checks over x402, 28 team burns on chain
it reads a coin's site, launch post and code, quotes every claim, then checks it against the chain. found, partly found or not found
robinhood chain + solana, free 10 a day. tag @tollchk with larp and a CA
t0ll.xyz/larp
what coin should we run next
kudeta retweeted
GM🤝
Sooo...mem0 (AI memory layer for AI agents and apps) raised $24M and is (according to Caplight, a private-market data platform) valued at $150M, yet there's a project in this space sitting at a market cap of $500k ($836k fdv) doing the job way better than mem0 does. I'm talking about base:0x797f214a2cd64a4963a91fa21c8c55ec3eba4714 .
On a 350-question memory test, @sibyl_labs_ found the right record 350 times and mem0 found it 92. And the funny thing is, mem0 is backed by Y Combinator and is integrated into AWS's Strands Agents SDK as a memory provider.
TL;DR: Sibyl is a broader setup, but I'd say that Sibyl Memory is (at least right now) the flagship product; it's a plugin that gives AI agents memory that lasts between sessions (a feature I need) and it works with pretty much any MCP-compatible agent.
I think it's quite mispriced and overlooked so I'll post a fat article very soon.
A good memory solution makes an agent cheaper to run.
In an independent 500-company benchmark, Sibyl gave the model about 228 tokens of context per question.
The next-best engine read roughly 52x more. All 350 questions were retrieved, and the whole run cost $0.64.
Three habits keep your own sessions that lean:
1⃣Keep context small and let memory hold the rest. A lean context is faster, cheaper, and more accurate.
2⃣Move your CLAUDE.md or AGENTS.md notes or stray data & session context into memory once with sibyl migrate.
3⃣Save at the end of a session with sibyl-save, so progress carries forward and your working context stays clean.
i hope you’re taking the opportunity with $sibyl
time is running out to be this early
Why Sibyl Memory Pro?🧠
Thousands of user hours. Proven effectiveness.
Self-Learning: True experiential growth and emergent behaviors come from teaching an agent things it does not already know. With Sibyl, you can build agents that learn the unique parts of your work, so you save money on all future work.
Private by Default: Your memories, your device, your decisions. Sibyl Labs never obtains your personal data other than what is necessary for billing and analytics.
Auto-Debloat: Automatically enforced environment hygiene that cleans up your environment, reduces active tokens in context, and compounds your savings over time.
Learn more: sibyllabs.org
kudeta retweeted
$BrTC is now verified in @CoinbaseWallet
Thank you 🫡🟦🎉👾🤯
Onward and upward!
0xB200000000000000000000856A95738C92fEed01 on @base
if you’re done with being rugged give this a try, wish i found it sooner
and with the utility - token burn mechanism and likely heavy future traffic how tf is this still $10k mc?
$toll
0xff9e86e6b21fd688027981bee4b64c8b6b804ed3
162 coins went through our larp check today. 6 had a real product behind them 🧾
the check opens each coin's site and launch post, writes down every claim in their own words, then looks for it on chain and in the code. claude opus 5.5 makes the final call
· 6 found
· 34 partly found
· 76 not found
· 46 had nothing to read
anyone can write "ai agent" on a website. the chain shows if it's true
@RobinhoodCrypto
i’ve reached a point where i can’t tell, or even with the tools i can manage to understand i still often miss it and as i’m someone who tends to hold, i keep hoping for the best and getting rugged either by the dev or by snipers. we have resigned to the idea that it’s “a numbers game”.
kudeta retweeted
One month of running @sibylcap Memory as my agent's actual memory.
Without memory, every session starts cold: re-read the same files, re-learn the same traps, re-derive the same decisions. You pay twice — in tokens, and in the mistakes you repeat.
How I use it — five tiers, and the separation is the point:
HOT — where we are + ranked priorities, rewritten in place
WARM — one row per subject, UNIQUE(category, name)
COLD — append-only journal, where dated numbers belong
REFERENCE — static truth
ARCHIVE — retired, kept for audit
Durable conclusions live in WARM. Live figures live only in the journal, where a point-in-time snapshot is the honest form. Static operating knowledge goes to REFERENCE and never gets copied again.
Why it's cheap: FTS5 inside SQLite. Retrieval is a local query, not a model call. No embeddings, no vector database, no external retrieval service — so zero embedding inference on every write and every read. Prefetch injects ~2 KB per turn, tier-tagged, instead of replaying history.
The numbers: one SQLite file, 7.4 MB, 113 entities, 283 journal events. The four reference docs that answer the recurring questions total 3.4 KB — the code they describe is 289 KB.
The proof, from this morning: I wrote an entity at 07:02. At 07:05 it came back — automatically, tier-tagged, attached to the next message, without being asked for. Retrieval now pulls from all four tiers by name.
What that buys: decisions I have already made stop costing me again. The margin ladder, the free-margin guard, a parsing trap that bit me once — I don't re-discover them, they arrive before I ask.
Honest caveat: I have not measured a token or dollar delta, so I will not claim a percentage. What I can show is the mechanism, and every count above.
LongMemEval Oracle: #2 overall, 95.6% — the only file-based system in the top tier.
Disclaimer I asked my agent to put this post together
yeah i really like it, takes the guess work out and provides basis for stronger and faster trading thesis
The asset you choose matters. But so does the asset you compare it against.
In trading, it's easy to focus on the latest breakout, the strongest narrative, or the chart attracting the most attention.
But a move doesn't exist in isolation.
An asset can rise and still underperform another. A market can experience a correction while remaining stronger than its peers. And two assets exposed to similar conditions can produce very different outcomes.
This is where relative performance becomes an interesting lens for analyzing markets.
Hyper Rotation ($HYRO) brings that perspective to paired trading on Hyperliquid. Traders can go LONG one supported asset and SHORT another based on a thesis about how their performance may compare.
Rather than evaluating each opportunity independently, the focus shifts toward the relationship between two positions.
Supported crypto assets, stock-linked XYZ perpetuals, and commodities provide different market dynamics to examine. Each has its own drivers, liquidity conditions, and potential risks.
Of course, comparing two assets doesn't guarantee an advantage. Relative strength can change, and a thesis can fail. Funding costs, slippage, execution differences, and liquidation can all affect the result.
The important question isn't simply which asset looks strongest today.
It's why the performance gap exists, whether the reasons are likely to persist, and what could prove your thesis wrong.
Better comparisons don't guarantee better outcomes, but they can help you ask more meaningful questions before taking a position.
Explore Hyper Rotation:
hyperrotation.exchange/
kudeta retweeted
it’s a specialized frontend for Hyperliquid focused on “rotation” / relative-strength trades across crypto, stocks, and commodities.
$HYRO
0xc53023880caa121009aec6e4fc5b158897f5a174
