Skip to content

Repository files navigation

Eva AI Assistant

Website | Documentation | Issues | License: MIT

Eva is a local-first AI assistant for conversation, durable memory, voice, automation, and software work. It brings cloud and local models into one focused desktop experience while keeping configuration, approvals, and personal data under your control.

Eva splash

Highlights

  • One intelligent gateway: route work across OpenAI, GitHub Copilot ACP, LM Studio, and MCP tools without changing the conversation model.
  • Durable, inspectable memory: retain explicit facts locally with provenance linked to source conversation turns, lifecycle controls, corrections, bounded last-session recall, and a dedicated Memory view.
  • Workspace control room: monitor independent terminals and coding agents for the selected workspace, focus a tile, interrupt work, or send new direction. Project files stay beside Live and Details tabs. Create local Git workspaces or import existing projects through a responsive in-app folder browser.
  • Local-first operation: use SQLite and local models by default, with cloud services enabled only when configured.

Quick Start

curl -fsSL https://appatalks.github.io/eva-agent/get-eva.sh | bash
eva

Eva is also added to the system application menu. On first launch, select Eva (AIG) for the integrated routing, memory, and tool experience.

Current development version: 5.6.11 (Eva Standalone-5.6.11.AppImage, unreleased). Latest published standalone release: Eva Standalone-5.6.10.AppImage.

Preview

Eva chat workspace Eva voice interface
Eva settings workspace Eva agent operations
Eva coding workspaces Eva skills and memory

Trust by Design

  • Explicit user facts are committed before Eva acknowledges saving them; ordinary questions do not become memory writes.
  • Consequential actions remain behind confirmation and authorization boundaries.
  • Live answers use source receipts and fail visibly when current data is unavailable.
  • Secrets, runtime state, and personal memory stay outside the repository and release artifacts.

Documentation

Releases

Used by

Contributors

Languages