Versions:
IronClaw 0.24.0, published by NEAR AI, is an open-source security-oriented framework designed to host and execute autonomous AI agents inside hardware-encrypted enclaves on the NEAR AI Cloud. Positioned as a privacy-first alternative within the Developer Tools / AI & Machine Learning category, the platform keeps every byte of user data local, encrypted, and beyond the reach of the underlying large-language model, ensuring that credentials, prompts, and intermediate results never leak to third parties. The architecture couples a WASM sandbox that isolates untrusted plug-ins through capability-based permissions with an orchestrator/worker Docker pattern that issues per-job tokens, while built-in prompt-injection detection, content sanitization, endpoint allow-listing, and automatic secret injection at the host boundary provide defense-in-depth against data exfiltration. Because the code base is fully auditable and telemetry-free, developers and organizations can expand agent capabilities on the fly by authoring new tools without waiting for vendor updates, then deploy them across multiple input channels—interactive REPL, HTTP webhooks, web gateway, or WASM channels such as Telegram and Slack—without sacrificing compliance or confidentiality. Since its first public appearance fourteen versions ago, IronClaw has evolved into a self-hosted solution suited for automating sensitive business workflows, managing personal productivity pipelines, and experimenting with trustworthy AI orchestration at scale. The software is available for free on get.nero.com, with downloads provided via trusted Windows package sources (e.g. winget), always delivering the latest version, and supporting batch installation of multiple applications.
Tags: