Aevra

•
4
•

Enables AI chat interfaces like ChatGPT and Claude to securely control local devices and servers as agents.

Aevra acts as a local, workspace-scoped execution gateway, empowering AI clients to interact with your device and server environment securely. It provides a robust framework that applies comprehensive session, workspace, capability, permission, risk, approval, recovery, and audit rules before any isolated local Worker executes commands or modifies files. With deliberately separated data and control planes, Aevra ensures that AI agents can remotely manage projects through their respective platforms while maintaining strict policy enforcement and sandboxed operations.

主な機能

01Secure remote access via Cloudflare Tunnel with OAuth and Bearer authentication
02Strict command sandboxing using Docker or Podman
031 GitHub stars
04Policy-driven execution gateway for AI clients
05Comprehensive change management with journaling, recovery, and rollback capabilities
06Integrated sensitive file and secret management with DLP and audit logging

ユースケース

01Enabling remote AI-driven code execution and file manipulation within a sandboxed environment
02Allowing AI agents to securely manage local development projects and server configurations
03Providing a controlled interface for AI to interact with Git repositories and managed processes