If you need a solid private AI stack for your Bitcoin books consider: Venice AI for natural-language inference. No prompt logging, no training on your data. Clams for the local accounting engine. Opencode as the agent shell. New guide - https://clams.tech/blog/venice-ai-bitcoin-accounting-setup-guide/ Need a guide for a different stack? Let us know.

We put together a comprehensive list of all Clams features on our site as we had some feedback. Some highlights: 1. Reconciles Lightning channels with on-chain activity 2. Auto-matches self-transfers, no double-counting 3. No caps on wallets, transactions, or reports 4. Drivable by scripts and agents via CLI 5. Runs fully local after sync View the full list here - https://clams.tech/features/

If you need to find your cost basis for the 1099-DA form, Clams will do this for you. Fully private, data never leaves your machine. No sharing addresses or xpub’s with anyone. No limits on wallets or transactions. We have guides on our YouTube channel and blog. And we are happy to help if you are stuck.

Clams v1.0.0-beta.10 is out. Historical balance sheets are now available. The report takes an as-of flag, so you can pick any point in the past and get the books as they stood that day. Useful for audits, year-end, board reports, and due diligence. Lightning sync reliability got better across the board. Core Lightning works more cleanly over Tor-backed Commando. Long syncs are much more stable. LND pending-invoice handling is more reliable for nodes carrying many open invoices. On-chain wallets are faster too, with quicker incremental syncs after the first scan and recovery from older checkpoint formats. Testnet4 is now supported across Address, Descriptor, and XPub connections. A new flag lets you force a full sync and rebuild from scratch when you need to. Run clams update. Full notes: https://clams.tech/blog/release-v1-0-0-beta-10/

Let us know if you need any support!

With 2 Lightning nodes connected, Clams shows a modest capital gain. I connect the xpub for an on-chain wallet and the River exchange data - It's now a capital loss. At the start of the demo, sats moving between our wallets appear to be spends and receives. Without the full picture, there's no way to know they're transfers. Adding the on-chain wallet and the River CSV lets Clams track the bitcoin end to end: River Exchange → on-chain wallet → lightning nodes. Disposals get recategorized as transfers. In other words, they are no longer taxable events. The average cost basis went up too. River has the actual purchase prices, not inferred ones from on-chain timestamps. Higher cost basis + transfers reclassified = realized gains flip from positive to negative. "But I don't want to hand over my xpub to get a more accurate cost basis." You don't. Your data never leaves your machine. You can connect your own Electrum, Esplora or Bitcoin RPC server for on-chain lookups. More connections in Clams doesn't mean more trust. It means fewer false taxable events and an accurate cost basis.