NearGems

140k mcap btw solana:4btUt5tQrN9MSfiHzZMdHf2ognjQZrWErjSJ9aiApump is near:native nearly.trade/singularty.near…

Sup everyone! 👋 Dropping my updated first open-source project A1 (Know Your Agent) 2.8.0 github.com/dyologician/A1 It's now not only for devs, even you can use it. It's now for builders and for your personal AI agent like Openclaw. github.com/dyologician/A1/wi… WHAT IS IT? A1 is a cryptographic identity and authorization layer specifically built for AI agents, especially when they're talking to each other, delegating tasks, or running as part of a larger system. Think of it like a passport system for AI agents. Every agent gets a signed identity, a passport, that says exactly what it can do, who authorized it, and for how long. When it hands off work to another agent, that sub-agent gets a signed certificate chained back to the original. Every action leaves a tamper-evident receipt. The whole thing verifies offline, no cloud call needed. WHAT'S IT FOR? You're running multi-agent AI systems, one agent researches, another executes trades, another manages files, and they're delegating to each other, sometimes several layers deep. Something goes wrong. Who authorized that? Not "which API key was used." Not "what does the log say." Which human said this specific agent was allowed to do this specific thing, right now, provably and cryptographically? Right now most multi-agent setups have no honest answer to that. A1 is built to fix that. WHAT PROBLEM DOES IT SOLVE? There's a gap in how multi-agent AI works today called the Recursive Delegation Gap. When Agent A delegates to Agent B which delegates to Agent C, existing tools like JWTs, OAuth, or SPIFFE completely fall apart. They weren't built for this. • JWTs don't chain, Agent A's token says nothing about Agent B • OAuth scopes are just strings, not cryptographic, not provable • Audit logs are mutable, they're a record, not a proof • No existing system can prove which human authorized the action at the bottom of a multi-agent chain A1 closes this with a chain of cryptographically signed delegation certificates. Every hop is signed. Every capability is enforced mathematically (not by policy). Every action produces a ProvableReceipt, an immutable proof anyone can verify independently, without secrets. WHAT CAN IT DO? • Agent Passports, give any AI agent a cryptographic identity that defines exactly what it's allowed to do • NarrowingMatrix, 256-bit capability enforcement via bitwise AND; a sub-agent mathematically cannot escalate beyond what its parent was granted, enforced at the compiler level with zero unsafe Rust • Provable Receipts, every authorized action generates a tamper-evident audit proof with a Blake3 commitment over the enforced capability mask • ReasoningTrace, agents can cryptographically commit their step-by-step reasoning (Thought → ToolCall → Observation → Decision → FinalAction) as a Merkle tree bound to the chain fingerprint; individual steps can be selectively disclosed to auditors without revealing the full trace • Governance & Approval Gates, specific capabilities can require a named human approver to sign off before the action runs; key rotation policies built in • Agent-to-Agent Negotiation, a full signed 3-way handshake protocol (CapabilityRequest → DelegationOffer → DelegationAcceptance) for agents to negotiate capabilities between themselves • On-chain Anchoring, anchor receipts on Ethereum, Polygon, Base, Arbitrum, and Solana • Fully offline verification, no network call, no trust server needed at authorization time • Per-principal rate limiting, built into the authorization layer • One-line integration, @a1_guard in Python, withA1Passport in TypeScript, a1.WithPassport in Go • Works with LangChain, LangGraph, LlamaIndex, AutoGen, CrewAI, Semantic Kernel, OpenAI Agents SDK • Built-in MCP server, Claude Code and Cursor work with zero code changes, just point at the gateway • Post-quantum hybrid signatures, ML-DSA + Ed25519, ready when you need it • W3C DID + Verifiable Credentials out of the box • Enterprise KMS backends, AWS KMS, GCP KMS, HashiCorp Vault, Azure Key Vault • SOC 2 and ISO 27001 compliance mappings included • A1 Studio, full web dashboard at /studio for managing passports, visualizing chains, and testing without touching a terminal First public repo, 🙏 github.com/dyologician/A1

@noir_wallet is actually cooking for $ZEC fam fr 👀 Proper privacy first wallet with real ZEC DeFi integration is finally here dawwwgg The first @Zcash Chrome Extension Wallet