Discover and explore top open-source AI tools and projects—updated daily.
AzmxAISovereign agent platform for local AI development
New!
Top 67.8% on SourcePulse
Summary
AZMX AI is a sovereign agent platform prioritizing privacy and local control for developers and power users. It enables AI agents to run directly on the user's machine, routing requests to local models or chosen providers without proxying through AZMX servers, enhancing security and workflow efficiency.
How It Works
The platform ensures privacy by processing AI requests locally, connecting directly to user-selected providers (OpenAI, Anthropic, Google, etc.) or local Ollama models. Key features include "per-call agent approval" for actions like file writes/shell commands (configurable security), and a hash-chained, tamper-evident local audit log. AZMX is MCP-native, integrating with numerous services (GitHub, Kubernetes, Postgres) via configurable JSON manifests.
Quick Start & Requirements
Installation is available via Homebrew (macOS), a curl script (macOS/Linux), or winget (Windows), with signed installers provided for all platforms. The core installer is ~10 MB with sub-second cold starts. Local AI setup requires Ollama (~150 MB download, ~4.7 GB for models). Detailed guides are in SETUP.md.
Highlighted Details
Maintenance & Community
Community contributions (skills, agents, MCP connectors, snippets) are encouraged, with contributors retaining copyright and granting permissive licenses. Contributions are integrated via auto-updater. Support for bugs is via GitHub issues.
Licensing & Compatibility
The core application is under an End-User License Agreement (EULA). Official npm packages (@azmxailabs/mcp, @azmxailabs/agent-sdk) are MIT-licensed. The platform supports air-gapped operation and requires no account or telemetry.
Limitations & Caveats
The core AZMX AI application source code is proprietary. The EULA governing the main application may contain usage restrictions not detailed in the README.
1 day ago
Inactive
e2b-dev