Mobile keyboards were not built for ssh user@host. AI agents were not built for 80-column terminals. Evern fixes both.
Conversation-First
Replace bash prompts with natural language. Speak or type what you want — Evern streams the AI response as rich, interactive cards instead of raw terminal output. Code diffs, file trees, and error traces render as first-class UI, not monospaced walls of text. The terminal is always one tap away when you need it.
Smart Rendering
A three-layer rendering pipeline transforms raw terminal output into structured, readable content. ANSI escape codes become styled text. Markdown and diffs become formatted cards. Agent actions become interactive UI components. Each layer builds on the last — nothing is lost, everything is gained.
Never Drops
Mosh protocol keeps your session alive through network transitions, dead zones, and IP changes. SSH reconnects automatically. Your work never stops.
Design Identities
Five built-in design identities, each with its own typography scale, spacing rhythm, and color palette. Switch between them or let the platform decide.
Architecture
| Core | Rust |
| Rendering | GPU-accelerated |
| Protocols | SSH2, Mosh |
| Platforms | iOS, Android |
| AI Layer | Agent-agnostic |
| License | Apache 2.0 |
A single Rust core powers both platforms. The terminal emulator, SSH client, Mosh implementation, and AI rendering pipeline are all shared code — native Swift UI on iOS, Kotlin on Android. GPU-accelerated rendering handles even the densest terminal output at 120fps.
The AI layer is agent-agnostic. Claude, GPT, Copilot, or local models — Evern detects and renders their output natively, regardless of which agent your server runs.
Free forever. No feature gates. No telemetry. No accounts. Apache 2.0.
Follow the Build
Evern is in active development. Star the repo to follow progress.