Give an AI agent a verifiable, encrypted memory trail on the X1 blockchain.
Preview and approve wallet transfers before an agent moves any funds.
Verify recovered encrypted memory against public on-chain hashes.
Register and check an agent's Agent ID Protocol identity before writing memory.
| syntharalabs/hxmp | 0xmukesh/docusaurus-tutorial | a15n/andrewscheuermann | |
|---|---|---|---|
| Stars | 1 | 1 | 1 |
| Language | JavaScript | JavaScript | JavaScript |
| Last pushed | — | 2021-12-27 | 2015-01-11 |
| Maintenance | — | Dormant | Dormant |
| Setup difficulty | hard | easy | moderate |
| Complexity | 5/5 | 2/5 | 1/5 |
| Audience | developer | developer | general |
Figures from each repo's GitHub metadata at analysis time.
Requires an X1 wallet funded with XNT and a verified Agent ID Protocol identity before memory writes work.
HXMP, short for Hermes X1 Memory Protocol, is a toolkit that lets AI agents store and retrieve memory on the X1 blockchain in a structured way. The idea is that agents get a paper trail of receipts and identity linked records on the public chain, while the actual content of that memory stays encrypted and readable only by whoever holds the private key. The project sums up its design goal as public proof paired with private readability. An agent using HXMP can verify its identity through a separate system called the Agent ID Protocol, create encrypted memory entries, publish hashes of those entries on chain without revealing their contents, and organize memory into named lanes with sequence numbers so it can page back through its own history. It can also recover and decrypt memory later, check that recovered memory matches the hashes stored on chain, and write plain non secret receipts for important actions it takes. Beyond memory, the tool can preview and send token transfers on X1, and interact with token creation and liquidity features, always with an explicit approval step first. The project is upfront about its limits: it does not hide activity from outside observers, who can still see wallet addresses, timestamps, and transaction records. It also warns strongly against ever storing private personal information, passwords, seed phrases, or other sensitive data through this system. Two separate keys are used, one for signing transactions and paying network fees, and a different one purely for encrypting and decrypting memory locally, and the signing key is never printed out. Any action that only reads data can run without special permission, but anything that changes state, such as writing new memory, registering an identity, or moving funds, requires the user to explicitly approve it first, often by confirming an exact preview hash before execution. The README describes this as early, experimental infrastructure, and advises treating any wallet used with it as a hot wallet holding only small, disposable amounts rather than real savings.
A toolkit letting AI agents store encrypted, identity linked memory and receipts on the X1 blockchain with explicit approval controls.
Mainly JavaScript. The stack also includes JavaScript, Node.js, X1 blockchain.
Setup difficulty is rated hard, with roughly 1h+ to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.