Getting started

Eyeban is a native desktop AI coding workbench paired with a cloud coding agent. The desktop app is the primary surface; the agent runs over the network against Eyeban cloud at api.eyeban.ir.

What you get in the workbench

The workbench surfaces the pieces you already expect from a coding environment, wired to the agent:

• Explorer for the project tree
• Search across the workspace
• Source control (SCM) for status, diffs, and commits
• Integrated terminal
• Code editor with local language-server completions
• Agent chat for multi-step coding work

Agent modes

Four modes cover different risk and workflow profiles:

• Agent — implement multi-file changes with tools
• Plan — design an approach before writing product code
• Ask — read-only exploration and answers
• Debug — investigate failures with evidence and reproduction before fixing

Switch between modes when the task changes shape. Guardrails differ by mode; see the Agent page for the full table.

Cloud, accounts, and keys

Authentication, model routing, entitlements, and agent runs go through api.eyeban.ir. Default provider keys stay on Eyeban servers. Optional bring-your-own-key proxies your key per request without storing it server-side. Your account dashboard shows current plan entitlements and credit/wallet state.

Version and availability

Current product version is 3.1.2. Windows Setup and the Linux .deb are on the Download page. The site and docs are bilingual (English and فارسی).

A sensible first hour

When you have a build, open a real repository, skim the tree and tests, then ask for a small change you can review end-to-end. Prefer Contact for enterprise evaluation or security reports.