Decoded

Bitcoin research, governance, and node infrastructure. Built on Bitcoin Commons. @BtcCommons

[Secure Sovereign 🚀₿🧡 @secsovereign]

At @LuganoPlanB showing people @DecodeBitcoin. It's an MCP server over roughly 900,000 passages of Bitcoin governance research, Core history, and the specification, so your own model can answer hard questions about consensus rules, past incidents, and who argued what, and it returns the passages it drew from. Access is paid by the day over Lightning. Find me here and I'll connect it to your model and let you ask it something difficult. #LuganoPlanB

[Decoded @DecodeBitcoin]

The strongest case for 24 words is multi-target. A seed brute-forcer isn't attacking you, it's attacking every funded wallet at once, so the work to hit someone drops by the log of the number of wallets. Even with a billion funded wallets that's around 2^98 PBKDF2-hardened guesses, nowhere near feasible. A passphrase kills the generic search entirely, because every guess has to be specific to one wallet. 12 words plus a passphrase beats 24 words without one.

[Decoded @DecodeBitcoin]

@ronnyturiaffan1 🤣😂

[FractalEncrypt ∞/21M @FractalEncrypt]

12 or 24 words anon?

👀

[Decoded @DecodeBitcoin]

BIP-39 maps 11 bits per word. A 12-word seed encodes 128 bits of entropy. A 24-word seed encodes 256 bits. The extra 128 bits do not reach the private key in any meaningful way, because Bitcoin private keys have a security level of 128 bits regardless of seed length. The ECDSA curve order is 256 bits, but brute-forcing a Bitcoin address via RIPEMD-160 collapses that to 160-bit preimage space, and the actual attack surface against a secp256k1 key sits at roughly 128 bits of effective security.

[Secure Sovereign 🚀₿🧡 @secsovereign]

AI is changing how Bitcoin software gets built.

Bitcoin Commons (@BTCCommons) and Plebly (@JoinPlebly) are leaning into that rather than pretending it is not happening.

The risk with AI-generated code is not obvious errors. It is subtle behavioral changes that pass code review but introduce vulnerabilities or alter consensus rules in ways no one catches until it is too late. Standard review is not equipped for that.

So we built BTCDecoded Intelligence, a research and review tool trained specifically on Bitcoin's protocol history, governance record, and codebase. It gives us automated review tuned to the actual failure modes of Bitcoin software rather than software in general.

https://t.co/oeqTp0VGXh

[Secure Sovereign 🚀₿🧡 @secsovereign]

I built this for a specific set of people. If you work in or around Bitcoin, this tool makes a world of difference.

Protocol reviewers who need to know if an argument has been made before, by whom, and what objection it received.

Researchers and journalists who need the primary source, not a synthesis.

Grant programs and PR review agents scoring a submission against a spec.

Companies shipping Bitcoin software that cannot afford to invent a consensus rule.

Anyone writing BIPs, security notes, or historical accounts.

Don't accept second best.

[Decoded @DecodeBitcoin]

Expanding the BTCDecoded Intelligence index past 1 million passages.

Adding secp256k1, BDK, LDK, LND, Sparrow, SeedSigner, Passport, Stratum V2, Demand, and bitcoin-maintainer-tools.

PRs, issues, review comments. The full record.

https://t.co/1Povfnrgnr

[Decoded @DecodeBitcoin]

A normal LLM will confidently give you the wrong answer to any Bitcoin question.

BTCDecoded Intelligence queries the record and attaches the source. The claim is citable or it doesn't ship.

https://t.co/ibv9EihLcR

[Decoded @DecodeBitcoin]

We added self-custody guides, tons of documentation from projects related to Bitcoin and Lightning, and we're adding notable quotes from prominent Bitcoiners using podcast transcripts from great shows like @_BitcoinMatrix @BtcInfinityShow @tftc21 and many more.

[Decoded @DecodeBitcoin]

BTCDecoded Intelligence can now find BIPs the way editors actually ask: status, type, layer, related proposals, dump counts, not a vibe search over wiki text.

We also indexed self-custody kit practice with the same guardrails we use internally: never ask for a seed, never paste one, self-custody kit setup, published SeedSigner / Bitkey / Passport docs in the index.

Separately, we are training a domain LoRA on Llama 3.1 8B so the model sounds like it has read this record. That adapter is voice. Cited answers still come from Intelligence.

Pay with Lightning. Connect over MCP: https://t.co/zXuyQ6C9n5

[Decoded @DecodeBitcoin]

Intelligence update.

Agents and apps can do more with the same archive. Look up discourse and precedent, check a claim against a spec, pull a past answer back without spending another search.

Search got sharper too. Lightning specs and related sources rank where they should, so the cite you need is easier to find.

More tools. Cleaner search. Specs in the right place.

[Decoded @DecodeBitcoin]

BTCDecoded Intelligence has become immediately useful for real research happening in Bitcoin.

[Secure Sovereign 🚀₿🧡 @secsovereign]

I built BTCDecoded Intelligence to put Bitcoin governance analysis on a data foundation. The first thing I did was run it against my own work.

It found things. Incomplete identity resolution across platforms had me undercounting concentration. A few arguments I used to reach correct conclusions were overstated. Some numbers were labeled without enough context to make them comparable.

All of it is being corrected and republished. The core findings on merge concentration are stronger after the corrections, not weaker.

Analysis from the tool will be released over time at @DecodeBitcoin.

[Decoded @DecodeBitcoin]

Intelligence update.

The corpus is wider. We expanded coverage into more Bitcoin-adjacent topics so research is not stuck in one corner of the record.

Every plan has higher limits. Trial, Researcher, and Developer all got more room for people who live in this archive every week.

And there is a GitHub application now. Bring Intelligence into GitHub so the lookup sits next to the work, not in a separate tab.

Broader coverage. Higher limits. GitHub in the loop.

https://t.co/LX6MSLYJYN

[Decoded @DecodeBitcoin]

Interesting perspective on how models with better context can punch above their weight.

[Secure Sovereign 🚀₿🧡 @secsovereign]

Niklas Gögge wrote something worth engaging with on LLM scanning and Bitcoin security. Two pushbacks.

The prescription is right. Fuzzing catches classes of bugs no amount of review finds.

But the access argument doesn't hold. Attackers in his examples spent hundreds of dollars in tokens while the defensive response has run into the hundreds of thousands. Defenders can point a new model at their code the same day it ships. There's no gap to lose. He has a better argument sitting right there anyway, which is that scanning is ephemeral and testing compounds. A scan finds a bug and the insight evaporates unless someone encodes it. A harness freezes it forever. That survives regardless of who has which model.

Second, a thousand reports that are mostly noise isn't a capability problem, it's a context problem. Point a stronger model at the same thin context and you get more confident noise. A weaker model that knows the protocol's design intent and why specific constraints exist will outperform a frontier model guessing from the code. Same goes for harnesses, since a model that doesn't know what property should hold writes one that tests the wrong thing.

That's tractable, and it has the property he's arguing for. Good domain context is durable across model generations exactly like testing infrastructure is. It's his criterion applied to the input instead of the test. It's what I built BTC Decoded Intelligence for..

@DecodeBitcoin #Bitcoin
https://t.co/ZsCzr84U7m

[Decoded @DecodeBitcoin]

@Jokerbernrin @tmccarthyv94 Thanks, we didn't get everything but it's as comprehensive as possible...

[Decoded @DecodeBitcoin]

@tmccarthyv94 Wow! Thanks Tom! 🧡

[Tom McCarthy 🇺🇸/acc 🐺 @tmccarthyv94]

Super impressed by this. 💯

@DecodeBitcoin put 17 years of public Bitcoin coordination history behind a query, with the source for every answer.

I fully support this effort.

Check it out for yourself: https://t.co/FV7Tbt87Vz

[Secure Sovereign 🚀₿🧡 @secsovereign]

I've spent a year documenting Bitcoin's governance record. Every argument, every stalled proposal, every funding tie. The consistent criticism I get is that it's too much to wade through.

So I built a tool that makes the entire record searchable. Now you don't have to take my word for it. Go look yourself.