Confidential, verifiable execution for crypto and AI
Joined April 2021
- Tweets3K
- Following50
- Followers84.2K
- Likes5.2K
Pinned Tweet
Your wallet’s security depends on who can change its rules. Hosted wallets like @privy_io and @turnkeyhq keep those rules inside their own system.
With Lit Protocol, the rules live in contracts on @base. Changing them takes a transaction from your wallet, and anyone can read them.
Teams like @RelayProtocol and @EmblemVault build with Lit for real control.
No single employee should be able to move a fintech's funds.
On Lit, the signing key stays in an enclave, every payment runs through code next to it, and in ChainSecured mode the contract only accepts rule changes from your own Safe.
Here is a payment rule that needs two approvers for large payments, and the setup that keeps any one person from changing it.
spark.litprotocol.com/no-sin…
Builders can’t be stopped!
Where does your app keep its private keys? If the honest answer makes you a little nervous, I'd love to hear from you.
I've been building on @LitProtocol for a month now and have shipped two projects already. If a day goes by without me building something, I feel like I skipped a meal 😅 For my next build, I want to solve a real problem, so I'd love to hear yours first.
Has something like this happened to you?
• A private key sitting in a .env file, or in a commit you had to scrub
• An AI agent or bot that needed API keys or a wallet, with nothing really stopping it from misusing them 🤖
• A payout or trading script where one bug could sign the wrong transaction
• A client asking who on your team can access their keys or data
Or something else entirely? Tell me what happened and what you did about it. Replies and DMs open📥
Encrypted memory protects an AI agent's data from the host. The agent's own outbound calls are a separate risk.
With Lit, the agent runs in a confidential VM, a proxy inside it allows only the hostnames you listed, and changing that list needs your approval on-chain.
Here is how to run an agent on customer data so the hosting provider can't read it, and what your customers can check.
spark.litprotocol.com/ai-age…
An AI agent that holds a production key can leak it into logs, transcripts, and prompts.
With Agent Keychain, the agent asks for one operation, like reading a Stripe balance. The agent never gets the key. It is decrypted only inside a Lit enclave to make that one call.
How it works, how to set it up for Claude Code, and what it does not protect against:
spark.litprotocol.com/ai-age…
Ethereum and Solana use different kinds of keys, so a wallet for both usually starts from one seed phrase.
On Lit, one action's key covers both. It works as an Ethereum key as it is, and the same 32 bytes work as a Solana seed.
Here is one action that signs on both chains, with a limit for each and the client code for Solana.
spark.litprotocol.com/wallet…
If your AI agent holds a private key, a prompt injection can spend from it.
A safer setup gives the agent a tool that asks for a payment. The key and the rules for each payment stay in an enclave where the agent cannot reach them.
We wrote up how to build that on Lit, with the rule, the tool definition, and the setup calls.
spark.litprotocol.com/give-y…
Auditors will ask you to prove that no employee can reach your customers' keys.
You can answer with a setup where no employee has a technical path to the key, plus checks the auditor can run without trusting you.
Here is how that works on Lit: keys derived in an attested enclave, rules owned by your own Safe on Base, the commands an auditor runs, and the limits to disclose.
spark.litprotocol.com/prove-…
You can run a crypto trading bot without a server and without a private key that anyone can copy.
The bot is a Lit Action that Lit Triggers runs every hour. It checks the price, follows limits written into its code, and signs with a key only that code can use.
Here is the whole bot, which sells a little ETH on Base when the price is high, along with the setup commands.
spark.litprotocol.com/crypto…
Cool build from the community for identity verification!
Scammers don't hack you. They pretend🎭.
Someone in your DMs says they're a recruiter, or an admin or mod in a project's Discord? Send them a ProveIt link, powered by @LitProtocol.
Try it (linktr.ee/getproveit):
github.com/heissally/ProveIt
You can use an AI vendor's product on sensitive company data while limiting where that data can go.
With Lit, the vendor's container runs in a confidential VM, it can only reach the services you approve, and you check the attestation before sending anything.
How it works, where the data goes after the model reads it, and what you can verify:
spark.litprotocol.com/ai-on-…
The best place for a private key is somewhere your app can ask for a signature without ever reading the key.
On a phone that is the iOS Keychain or the Android Keystore. On a server it is a KMS or a signer your app calls.
This post covers where keys usually leak, what a KMS checks before it signs, and how to put an allowlist and a limit next to the key on Lit.
spark.litprotocol.com/where-…