Exchange Ranking
Top Exchanges
24h Volume Ranking
Popularity Ranking
Exchange BTC Balance
Proof of Reserves
Decentralized Exchanges
Funding Rate
Funding Heatmap
Liquidation Data
Max Pain
Long/Short Ratio
Whale L/S Ratio
Binance/Okex/Huobi L/S
Bitfinex Margin L/S
ETF Tracker
Solana ETF
XRP ETF
Hong Kong ETF
Bitcoin Treasuries
Crypto Reversal
Ethereum Reserves
HyperLiquid Wallet Analysis
Hyperliquid Whale Watch
Large Transactions
On-chain Movement
Bitcoin ROI
Stablecoin Market Cap
Options Analysis
News
Articles
Economic Calendar
Features
Wallet
Contract Calculator
Security
Collections
Watchlist
Following
NEAR NEAR Protocol AI releases IronClaw version 1.2: Enhance security and multi-user support
NEAR Protocol AI releases IronClaw AI version 1.2. As its open source security proxy platform, this update expands runtime capabilities, adds multi-user support, broader tool connectivity, and strengthens defense against prompt injection attacks.
Open tool connectivity via MCP
The main highlight of version 1.2 is comprehensive support for any MCP-compatible tool, covering everything from CRM systems to internal APIs. IronClaw's Model Context Protocol (MCP) integration allows users to connect to the MCP server for additional functionality. Its plug-in architecture supports adding new WASM tools and channels directly without restarting the agent. This flexibility allows teams to connect agents to existing enterprise tools without having to rebuild the technology stack.
Agents can now run directly on Slack channels and Telegram groups, and each message is processed based on the identity, permissions and approval thresholds of the sending user. IronClaw supports multi-channel access through Slack, Telegram and Web Gateway, as well as background tasks, heartbeats based monitoring and parallel work capabilities.
Security Model: Untrusted History and Encrypted Safe Areas
The update also strengthens IronClaw's way of handling potentially malicious content on shared channels. Channel history is treated as untrustworthy content rather than instructions; when agents read public channels, the platform protects against prompt injection attacks. IronClaw imposes strict restrictions on resource use and uses advanced safeguards against prompt injection.
IronClaw is built on Rust and deployed in a NEAR AI Cloud encrypted Trusted Execution Environment (TEE), designed to provide built-in privacy protection for autonomous agents and control access to user assets and data. The key is that user confidential information is isolated outside the model-API keys, tokens, and passwords are stored in cryptographic vaults, and IronClaw injects them only when the user authorizes them, and the AI never touches the original values.
All data is stored in the user's local PostgreSQL database and encrypted using AES-256-GCM. IronClaw also maintains a complete audit log of all tool activity and does not collect any telemetry data to ensure that data does not leave the user's local environment.
The core concept of this security design is very clear: AI agents are powerful, but if they lack security guarantees, being powerful will become a burden. These systems, which can hold credentials, browse the web and perform operations on behalf of users, can become threats to privacy, data and trust if not protected. IronClaw's architecture is built around this concern, ensuring that execution is verifiable and confidential information is protected at the hardware level.
Co-founded by Illia Polosukhin, NEAR Protocol has evolved from an Ethereum-like smart contract platform to a protocol that prioritizes AI integration and data privacy. Polosukhin is also co-author of the groundbreaking paper "Attention Is All You Need", which proposed the Transformer model.
